Research Article A Sentiment Delivering Estimate Scheme

Research ArticleA Sentiment Delivering Estimate Scheme Based onTrust Chain in Mobile Social Network

Meizi Li12 Yang Xiang1 Bo Zhang2 and Zhenhua Huang1

1College of Electronics and Information Engineering Tongji University Shanghai 200092 China2College of Information Mechanical and Electrical Engineering Shanghai Normal University Shanghai 200234 China

Correspondence should be addressed to Bo Zhang shzhangbogmailcom

Received 5 July 2015 Revised 20 September 2015 Accepted 27 September 2015

Academic Editor Jose Juan Pazos-Arias

Copyright copy 2015 Meizi Li et al This is an open access article distributed under the Creative Commons Attribution License whichpermits unrestricted use distribution and reproduction in any medium provided the original work is properly cited

User sentiment analysis has become a flourishing frontier in data mining mobile social network platform since the mobile socialnetwork plays a significant role in usersrsquo daily communication and sentiment interactionThis study studies the scheme of sentimentestimate by using the usersrsquo trustworthy relationships for evaluating sentiment delivering First we address an overview of sentimentdelivering estimate scheme and propose its related definitions that is trust chain among users sentiment semantics and sentimentontology Second this study proposes the trust chain model and its evaluation method which is composed of evaluation of atomicserial parallel and combined trust chainsThen we propose sentimentmodelingmethod by presenting its modeling rules Furtherwe propose the sentiment delivering estimate scheme from two aspects explicit and implicit sentiment delivering estimate schemesbased on trust chain and sentiment modeling method Finally examinations and results are given to further explain effectivenessand feasibility of our scheme

1 Introduction

Recently mobile social network has become one of the mostpopular platforms for people to daily communicate and sharetheir information [1ndash3] It allows users to post brief texts andexpress their opinions and then broadcasts them in publicSince the social network platforms are tied in mobile intelli-gent terminals people can communicate through mobility ofmobile social network platforms anywhere at any time withthem Such mobility of social network applications poses anethereal new dimension that overcomes the limitations oftime and space That is people can express their sentimentsin mobile social network more freely than in other tradi-tional media platformsWith the fission propagation patternsentiments would have larger and wider impacts in mobilesocial network such as Twitter [2] and SinaWeibo [3] mobilesocial network For example a famous and reputationaluserrsquos sentiment would have giant influences to hisher fansand followers Furthermore such impacts can be deliveredthrough the relationship networks of the fans and bringwider and deeper impacts in mobile social network Thatmeans sentiment influence in mobile social network is not

limited to direct impact but is more important on the indirectdiffusion impacts among users Under such explosion ofpublic sentiment there has been growing attention on theproblem of sentiment analysis in mobile social network

Traditionally sentiment analysis studies have focused onmining sentiment from individual texts and the evaluationof their impact through usersrsquo direct relationships [4ndash6]However as sentiment is propagated and interacted mutuallythrough user relationships which is a significant mannerof information and sentiment communication on socialnetwork platforms we consider that this feature of sentimentdelivering is essential and indispensable and cannot beignored for the study of sentiment analysis in social networksThus finding a way to estimate the sentiment deliveringmechanism and evaluate the impact of such sentimentdelivering is crucial for sentiment analysis in mobile socialnetworks which is the main goal of this paper

In this work there are two aspects which are takeninto account for studying sentiment analysis (1) sentimentin mobile social network including explicit sentiment andimplicit sentiment can be delivered from one to anotherthrough user relationship and (2) the key manner which

Hindawi Publishing CorporationMobile Information SystemsVolume 2015 Article ID 745095 20 pageshttpdxdoiorg1011552015745095

2 Mobile Information Systems

drives sentiment delivering is the trustworthiness amongusersThus by employing trust as the basis for our sentimentdelivering evaluation ourmain contributions in this work areas follows (1) proposing an overview of sentiment deliveringevaluation scheme is addressed and its related definitionsincluding trust chain sentiment semantics and sentimentontology are proposed (2) presenting a formal trust chainmodel which links users and measures the trustworthinessis proposed (3) presenting a sentiment modeling methodis presented based on a set of rules which enable sentimentsemantic to be calculated through definitions in sentimentontology and (4) proposing a sentiment delivering estimatescheme is therefore addressed which is composed of explicitand implicit sentiment estimate based on trust chain andsentiment modeling

The rest of the paper is organized as follows Section 2discusses related work of our study Section 3 presents anoverview of scheme and its related definitions Section 4addresses the trust chainmodel and its trust value calculationmethod Sections 5 and 6 present the modeling method anddelivering estimate schemes of explicit sentiment and implicitsentiment respectively Empirical results and discussion aregiven in Section 7 Finally Section 8 concludes the paper

2 Related Work

21 Trust Computation Trust as an inherent willingness ofhuman beings shows the emotional and rational confidencein between people It is derived from judging trustworthinessby evaluating various facts which can lead to either the con-fidence or distrust Over the past few years many works havefocused on computing trusted paths in network environmentsuch as P2P and mobile social networks (SNS) [6ndash10] Intrust evaluation there are two core kinds of trust as directtrust and indirect trust Direct trust is used for reflectingthe trustworthiness between direct connected users whileindirect trust is used widely in long path connected usersthrough intermediate users [6] Typically many historicaldata based on direct interactions such as behaviors com-ments or other various evidences are used to calculate thedirect trust degree among users [6] In most cases the directtrust reflects the subjective trustworthy feelings towards thetargets through the historical direct interactions See-To andHo [7] propose a method to evaluate the influence of truston oral comment in mobile social network Wu and Chiclana[8] present an approach to computing trust that users have allagreed on in mobile social network Wang and Gui [9] selecttransaction nodes in mobile social network and computetrust between them Based on real-time content Li et al [10]develop a community information recommendation systemwhich takes advantage of user experience Peiyun et al [11]propose an algorithm for computing direct trust value andfurthermore develop a Web service trust model based onmobile social network dynamic feedback Qiao et al [12]propose a context-based trust computation method wherethe trust generation principle in the psychology is employedIn addition the indirect trust is to evaluate trustworthinessthrough a deliverable perspective of trust concerning theindirect connected relationships among users [13ndash15] From

such point direct trust evaluation is also the foundationof indirect trust evaluation In most cases the calculationof direct and indirect trust is discussed based on a graphmodel direct trust can be seen as a direct connected edgebetween vertices while the indirect trust is seen as a longpath composed of direct trust in graph model Therefore thepath among indirect connected users is an essential factor forindirect trust evaluation Javier Ortega et al [13] propose amethod to compute a ranking of the users in a mobile socialnetwork and propagate both positive andnegative opinions ofthe users Then the opinions from each user about others caninfluence their global trust score including direct and indi-rect trust views Qureshi et al [14] propose a decentralizedframework and the related algorithms for trusted informationexchange and social interaction among users based on thedynamicity aware graph relabeling systemTheBellman-Fordalgorithm computes trust based on direct witness interactiontrust judgments [15] It generates a trust graph on the basis ofthe trust link between two peers who have direct interactionEach peer can submit or renew their trust judgments of othersbased on new direct interactions Further the trust betweenpeers is constantly updated by compounding old and newtrust judgments In addition the algorithm admits the mosttrustable path for trust computation it deems a long pathto be untrustworthy Golbeck proposed TidalTrust that getstrust in mobile social networks using numeric trust values[16] It utilized the shortest path based on the breadth-firstsearch Further TidalTrust can be used to retrieve accurateinformation from the highest trust adjacent nodes

However most of these above methods focus on eitherdirect trust or reputation computation and do not take intoaccount the chain relationship of trust among users andits complex path topology Different from traditional trustmodules there are the following considerations in this study(1) since the sentiments among users are considered to bedelivered through the user relationships the proposed trustmodel in this study therefore takes chain relationships amongusers into account to reflect the indirect confidencewhich canbe used to measure the probability of sentiment deliveringthrough users by their mutual trustworthiness (2) for directtrust calculation that is atomic trust chain in this workwe also use the weighted average method to calculate themwhich is similar to other traditional works However weintroduce a new factor of community which widely existsin mobile social network for weights evaluation That is thedirect trust calculation is impacted by the facts whether theusers are in same community or not and (3) with respect toindirect trust the proposed trust model calculates it throughuser indirect relationships The relationship compositionamong users that is the path topological information amongusers in social network graph is complex and variousUnfortunately few attentions have been paid to the complexpath composition for indirect trust That is there are manydifferent accessible paths between users and each accessiblepath manifests a delivering confidence Then the proposedtrust model aims to calculate the trust chain with taking allpaths confidence into account comprehensively in mobilesocial network which is also one of the main motivationswhich we work in this study

Mobile Information Systems 3

22 Sentiment Analysis Sentiment analysis has been exten-sively studied in recent years and various methodologicalschemes have been proposed for sentiment extraction inWeb image text or other areas [3 4] Sentiment analysishas been utilized in applications such as news tracking andsummarizing online forums file sharing chatting roomsand blogging [17] There are two typical classes for sen-timent analysis as machine learning based approach andsemantic based approach [17] In machine learning basedapproach a mass of material with real emotions are providedto machine for training machine to learn the rule s ofunderstanding and detecting sentiment [4 17ndash19] such assupport vector machine (SVM) k-means method NaiveBayesian (NB) LDA and Max Entropy (ME) In semanticbased approaches semantic analysis related methods areutilized for understanding the characters of sentiment [5 20ndash24] Typical methods of semantic based approach includenatural language processing method (NLP) keyword basedmethod emotion dictionary based method ontology basedmethod and computational linguistics for sentiment analysisIn the above semantic based sentiment analysis methodsknowledge about sentiment classification and representationis essential and indispensable for detecting and identifyingsentiment In addition some combined methods which usethe above machine learning and semantic analysis basedmethod together are proposed for sentiment analysis toimprove the precision of results [25] Besides explicit sen-timent mining implicit sentiment analysis has been paidmany attentions in mobile social network [26ndash28] Su et alpropose a mutual reinforcement principle which is basedon opinion evaluation through product feature words andopinion words to mine the hidden sentiment associations[26] Athar and Teufel explore methods to automaticallyidentify all mentions of a paper and then classify explicit andimplicit mentions for sentiment detection [27] Balahur et alpresent an approach towards automatically detecting implicitsentiment from contexts in which no clues of sentimentappear based on a commonsense knowledge base (namedEmotiNet) [28 29] However all thesemethods are still basedon existingmaterials and cannot estimate implicit sentimentsthrough a delivering view

In mobile social network sentiment has been extractedfor further analyzing in different platforms such as Twitter[23 30 31] MySpace [32] Facebook [33] YouTube [34]and other platforms [26 35 36] However most works ofsentiment analysis focus on detecting and mining explicitsentiments from objects of texts videos or audios In ourconsideration sentiment analysis should be studied in a viewof its dynamic characteristic Different from other worksthere are the following considerations for sentiment analysisin our study (1) the communication nature of mobile socialnetwork enables users to interact and share their opinions andemotions freely in the platforms whichmeans that sentimentinteractions with the effect of interactive influence are one ofthe most significant functions of mobile social network (2)sentiment can be delivered through users relationships andthen how to model the delivering process and evaluate thedegree of delivering is a crucial challenge and (3) besidesexplicit sentiment there are many implicit sentiments which

are kept by users and are not expressed explicitly and thenthese implicit sentiments should be recognized in sentimentanalysis

3 Overview and Related Definitions

The guiding principles we take in this work are as follows(1) there are two types of sentiment in mobile social networkas explicit sentiment and implicit sentiment (2) explicitsentiment which is expressed in users posts commentsor other ways directly can be modeled through a formalformat and calculated for its delivering estimate (3) implicitsentiment which has not been expressed visibly in mobilesocial network texts can be estimated by explicit sentimentdelivered through users trustworthy relationships

Therefore we address our scheme in mobile social net-work based on the trust chains among users The proposedscheme in this study is composed of three phases as (1) atrust chain establishing sentiment modeling and sentimentdelivering estimate As shown in Figure 1(a) a networkmodelof mobile social network is constructed based on graphtheory In such graph users are regarded as vertices whiletheir relationships are edges Then we can see that therewould be trust chains formed by usersrsquo past interactions andrelated data in social graph which is shown in Figure 1(b)Based on trust chains sentiments toward topic of users canthereby be delivered in mobile social network which revealthe impacts of sentiment in explicit and implicit ways InFigure 1(c) the explicit sentiment which comes from user Acan be delivered to B and C (in green solid lines) throughthe trust chain C rarr B rarr A (in blue solid lines) Andthen such influential explicit sentiment would in turn induceothers to express their different sentiments (in red solid lines)Likewise Figure 1(d) shows an example of implicit sentimentdelivering among users Due to the fact that users of D andE have not expressed their sentiments toward a topic inexplicit way we can evaluate the implicit sentiments of themaccording to explicit sentiment fromAbecause there is a trustchain E rarr D rarr A (in blue solid lines) between them Theexplicit sentiment of A can be delivered to D and E in implicitway (in green dot lines) and then generate implicit sentimentsof D and E towards the topic (in red dot lines)

Here we first address related definitions in this work

Definition 1 Trust chain model can be defined as Ω =

(119873 119865 119862 119879) where 119873 denotes nonempty set of user nodesin trust chain and the user nodes can be divided into threeroles as source user nodes 119873119878 intermediate user nodes 119873119868and target user nodes 119873119864

119865 sube 119873 times 119873 denotes the finiteset of atomic trust chain 119862 sube 119865 times 119865 cup 119865 119865 denotesthen combined trust chain which is composed of atomic trustchain and symbols of times and denotes serial trust chain andparallel trust chain respectively119879 119865 rarr [0 1]cup119862 rarr [0 1]

denotes trust value of atomic trust chain or combined trustchain

Through Definition 1 we can describe the direct andindirect trust relationships between users according to thetopology of chain combination formally


User relationshipUser

middot middot middot



(a) Original users and their rela-tions in microblog

UserTrust chain

(b) Trust chains amongusers

Topic

UserTrust chainExplicit sentimentExplicit sentiment delivering

A B C

(c) Explicit sentiment deliveringbased on trust chain in social net-work

Implicit sentimentImplicit sentiment deliveringTopic

UserTrust chainExplicit sentiment

A D E

(d) Implicit sentiment deliveringbased on trust chain in social net-work

Figure 1 Example of sentiment delivering based on trust chain in mobile social network

Definition 2 Sentiment in mobile social network is theopinion which user keeps to a topic or an event and canbe defined as 119878119864 = 119904119890

1 1199041198902 in which 119904119890

119894denotes one

dimension of sentiment of user

From Definition 2 we can see that the sentiment inthis work is modeled as a multiple-dimension entity todescribe userrsquos diverse feelings In the set of 119878119864 we definedthat each dimension of sentiment is a 3-tuple of 119904119890

119894=

⟨119905119910119901119890 119877119890119897119886119905119890119889 119897119890V119890119897⟩ where 119905119910119901119890 is the sentiment type119877119890119897119886119905119890119889 is the set of sentiment which includes related sen-timent and the calculating rules 119897119890V119890119897 isin [0 1] is the strongdegree of sentiment

For example a user 119906119894 has hisher sentiment semanticsas 119878119864(119906119894) = 1199041198901 119889119890119897119894119892ℎ119905 1199041198902 119908119900119903119903119910 which includestwo sentiment dimensions as 1199041198901 119889119890119897119894119892ℎ119905 and 1199041198902 119908119900119903119903119910According to Definition 2 the two sentiment dimensions canbe formally described as follows

1199041198901 119889119890119897119894119892ℎ119905 = ⟨119905119910119901119890 ℎ119886119901119901119894119899119890119904119904 119877119890119897119886119905119890119889

119895119900119910119891119906119897 | ⟨SMRTAR⟩ 119890119888119904119905119886119904119910 | ⟨SMRTMR⟩

119897119890V119890119897 09⟩

1199041198902 119908119900119903119903119910 = ⟨119905119910119901119890 119891119890119886119903 119877119890119897119886119905119890119889

119904119886119889119899119890119904119904 | ⟨SMRTAR⟩ ℎ119886119905119890 | ⟨TMR⟩ 119897119890V119890119897 03⟩

(1)

where the type of 1199041198901 119889119890119897119894119892ℎ119905 is happiness related sentiments

are 119895119900119910119891119906119897with calculating rules of SMRandTARand 119890119888119904119905119886119904119910with calculating rules of SMR and TMR the level value of1199041198901 119889119890119897119894119892ℎ119905 is 09 Likewise the type of 1199041198902 119908119900119903119903119910 isfear related sentiments are 119904119886119889119899119890119904119904 with calculating rulesof SMR and TAR and ℎ119886119905119890 with calculating rules of TMRthe level value of 119904119890

2 119908119900119903119903119910 is 03 Then the sentiment

dimension calculation can be executed based on given rulessuch as SMR TAR or TMR in the formal definitions Detailsof calculation rules are discussed in later sections

Similar to the above example sentiment can reflect userrsquosmultiple-dimension feeling or emotion focusing on a specificevent or topic It is apparently easy to evaluate onersquos explicitsentiment according to his explicit posting text behaviorscomments or other direct evidences For example if a userapproved posted positive comments followed the topicand forwarded posts to a topic in a near past he mighthighly keep a positive sentiment about the topic Howevermany users do not express their sentiments through theirdirect evidences These potential sentiments named implicitsentiment in this study cannot be measured directly basedon their past data But in our view sentiment is deliveredthrough usersrsquo relationships That is we can estimate implicitsentiment through relationships among users For example auser A who has no direct evidence to express his sentimentto a topic keeps a very high trust with his friend B whokeeps a strong positive sentiment to the topic In such contextwe can estimate that A might be likely to keep positivefeeling because his trustworthy friend B does so In thisexample implicit sentiment can be delivered through usersrsquotrust relationship which is also used for implicit sentimentestimation in this work

From above consideration we define two kinds of sen-timent as explicit sentiment and implicit sentiment whichis reflected through usersrsquo explicit and implicit behaviorsand evidences respectively Correspondingly the calculationmethods are based on the following rules explicit sentimentdegree is measured according to the direct sentiment evi-dences in past implicit sentiment degree is calculated basedon an estimate of trust chain delivery

Definition 3 Sentiment ontology is defined to describe senti-ment in a formal and normalized way for enabling sentimentto be understood by system automatically Sentiment ontol-ogy is a 3-tuple as 119878119874 = ⟨119862119897119886119904119904 119877119890119897119886119905119894119900119899 119877119906119897119890⟩ where 119862119897119886119904119904is a set of sentiment classes119877119890119897119886119905119894119900119899 is the set of relationshipsamong sentiment classes and119877119906119897119890 is the set of representationand calculation rules for sentiment


For example there is a subpart of sentiment ontology asfollows

119878119874 = ⟨119862119897119886119904119904 119877119900119900119905 ℎ119886119901119901119894119899119890119904119904 119904119886119889119899119890119904119904 119886119899119892119890119903

119877119890119897119886119905119894119900119899 119901119886119903119890119899119905-119888ℎ119894119897119889 ⟨119877119900119900119905 ℎ119886119901119901119894119899119890119904119904⟩

119901119886119903119890119899119905-119888ℎ119894119897119889 ⟨119877119900119900119905 119904119886119889119899119890119904119904⟩

119901119886119903119890119899119905-119888ℎ119894119897119889 ⟨119877119900119900119905 119886119899119892119890119903⟩ 119877119906119897119890 CMR SMR

TMRTARMER⟩

(2)

where the subpart of sentiment ontology contains fourclasses as Root happiness sadness and anger the classes ofhappiness sadness and anger are all the direct children ofRoot according to the 119901119886119903119890119899119905-119888ℎ119894119897119889 relation description thatis happiness sadness and anger are brothers in ontology andthe calculation rules in sentiment ontology are five rules givenin later section

Here we have the following related explanations forsentiment ontology

(1) Sentiment ontology describes sentiment in a treestructured model In such tree model each noneleafclass has just one parent and at least one child whileeach leaf class has just one parent and no child

(2) Each class has its layer number (119897119886119910119890119903) which indi-cates the distance from the class to the root of the tree

(3) There is a mapping function between sentimentdimension and sentiment ontology for describing thesemantic relation as 119891 119904119890

119894119905119910119901119890 rarr 119878119874119888119897119886119904119904

119895

(4) The set of 119877119906119897119890 in ontology aims to provide specialcalculation rules which can be used in sentimentmigration or changing The rule can be describedas 119903(119888119897119886119904119904

119894 119888119897119886119904119904

119895) = 119888119897119886119904119904

119896 which means that two

classes 119888119897119886119904119904119894and 119888119897119886119904119904

119895 can be changed into a new

class 119888119897119886119904119904119896 under the above rule 119903(119888119897119886119904119904

119894 119888119897119886119904119904

119895)

For the rules in ontology we here have an example for itscalculation as follows

SMR 119903 (119905119890119903119903119891119894119890119889 119908119900119903119903119910) = 119889119894119904119905119906119903119887119890119889 (3)

where the ontology defines a sentiment migration rule forcalculating two classes terrified and worry and then gets anew class disturbed Also there are additional conditions forthe above calculating rule and we will discuss them in latersection

Figure 2 shows an example of sentiment ontologywith theabove related explanations and its mapping with sentimentdimensions In this example we can see that the layer numberof class 6 is 2 and it has a mapping function with sentiment

1199041198903 Also we can get the layer numbers and semantic relations

as119897119886119910119890119903 (1198881198971198861199041199042) = 1

119897119886119910119890119903 (1198881198971198861199041199045) = 2

119897119886119910119890119903 (1198881198971198861199041199048) = 3

119891 1199041198901119905119910119901119890 997888rarr 1198781198741198881198971198861199041199042

119891 1199041198902119905119910119901119890 997888rarr 1198781198741198881198971198861199041199045

119891 1199041198904119905119910119901119890 997888rarr 1198781198741198881198971198861199041199048

(4)

Sentiment ontology can provide logic basis for sentimentevaluation andmake the evaluation reasonable on knowledgelayer In this study we assume that all sentiments can bedescribed by sentiment ontology Therefore userrsquos explicitsentiment can be modeled while the initial dimensions ofsentiment are extracted from userrsquos text comments andbehaviors in mobile social network Notice that in this studythework of analyzing initial dimensions of sentiment is not inour consideration andwould be discussed in our other works

4 Trust Measurement of Trust Chain Model

Here we propose the model of trust chains in detail basedon their different network topologies and their trust valuecalculation methods According to Definition 1 and the com-position method of trust chain we divide trust chain intofour cases according to the path composition as atomic serialparallel and combined trust chains

41 Atomic Trust Chain In atomic trust chain there is nointermediate node between two nodes That means the trustvalue is a kind of direct trust from one node to anotherFor our proposed atomic trust chain we also use a weightedaverage method by past historical data which is widely usedin many existing trust models However different from theother existing trust model we here introduce a new fact ofcommunity for weight evaluation in trust calculation Thefactor of community is not included inmost traditional directtrust computation methods since community is a specificentity in social network environment It is very commonthat users locate in community of SNS environment and theidentification of community implies onersquos trustworthiness tosome degreeTherefore trustworthiness level towards a userrsquoscommunity is crucial for direct trust calculation includingtwo cases two users in the same community and two users indifferent communities The former means that users have thesame community identifications and their atomic trust chainis established based on similar past experiences directly whilethe latter means that users have different identifications andtheir atomic trust chain is derived from the subjective trust ofthe whole group which the other belongs to

Therefore based on past historical data trust value ofatomic trust chain can be directly calculated by mutualinteraction records relationships and community identitiesin atomic trust chain Let there be two nodes 1198731

1198732and

community set 1198661 1198662((1198731isin 1198661) and (119873

2isin 1198662)) 119891119894(1198731 1198732)


Level(se4) = 1

Level(se3) = 08

Level(se2) = 08

1Level(se ) = 09

se4

se3

se2

se1

dimension and sentiment ontologyMapping function between sentiment

Sentiment dimension

Relation between classesClass in sentiment ontology

Class 9Class 8Class 7

Class 6Class 5Class 4Class 3

Class 2Class 1

Root

Figure 2 An example of sentiment ontology and its mapping with sentiment dimensions

denotes judgment which is delivered from 1198731to 1198732 We

denote119879(1198731 1198732) as trust value of atomic trust chain from119873

1

to1198732after119898 times of judgment And then 119879(119873

1 1198732) can be

calculated as follows

119879 (1198731 1198732) =

0 119899 = 0

sum

119899

119894=1(119891119894(1198731 1198732) times 119896119894)

119899

119899 ge 1

(5)

In the above equation we give the impact factor 119896119894 which

denotes impact degree for each judgment 119891119894(1198731 1198732) to the

accumulated trust value 119879(119873119894 119873119895)

119896119894can be calculated as follows

119896119894 =

1 evaluation 119891119894(1198731 1198732) happened in the same community

119890minus(1minus|119866

1cap1198662||1198661cup1198662|) evaluation119891

119894(1198731 1198732) happened in the different communities

(6)

42 Serial Trust Chain Serial trust chain means that therewould be a serial path from source node to target node andthe path has the following features (1) for source node itsout-degree is 1 and in-degree is 0 (2) for the target node itsout-degree is 0 and in-degree is 1 (3) for each intermediatenode its out-degree is 1 and in-degree is 1Thereby serial trustchain can be considered as the composition of atomic trustchain and the trust is transmitted one by one in such chainso that there would be an indirect trust from source node totarget node

Then we propose the value calculation method for serialtrust chain In our trust model serial trust chain reflectsan indirect trust among users We introduce the principleof Bellman-Ford algorithm [15] which deems a long pathto be untrustworthy for our proposed model of indirecttrust calculation That is the trust value of serial trust chainwould be decreased with the depth of trust chain increasingLikewise such rational is also applied for indirect trust ofparallel trust chain and combined trust chain

Let there be source node 119873119878 target node 119873

119864 and

intermediate node 119873119896

119868in serial trust chain and 119879(119873119894 119873119895)

which denotes trust value of atomic trust chain in the serialtrust chainTherefore we can calculate the trust value of serialtrust chain as follows

119879119862(119873119878 119873119864) =

1

119889119890119901119905ℎ (119873119864)

[119879 (119873119878 119873

1

119868)

119889119890119901119905ℎ(119873119864)

+

119889119890119901119905ℎ(119873119864)minus1

sum

119896=2

119879 (119873

119896

119868 119873

119896+1

119868)

119889119890119901119905ℎ(119873119864)minus119896+1

+ 119879 (119873

119889119890119901119905ℎ(119873119864)minus1

119868 119873119864)]

(7)

Here function 119889119890119901119905ℎ(119873119864) denotes the depth of serial trust

chain namely 119889119890119901119905ℎ(119873119864) = |Ω sdot 119873

119868| + 1 We can see that

the deeper the depth of trust chain is the weaker the trustvalue among users isThat means longer trust chain would be


punished due to the fact that trust can dampwith the numberincreasing of intermediate node

43 Parallel Trust Chain It is likely that there are two ormoretrust paths from source node to target node and there is nointersection node among the paths In that case we call thetrust chain as parallel trust chain That means parallel trustchain can be considered as composition of several serial trustchains between same source node and target node Likewiseparallel trust chain has the following features (1) for sourcenode its out-degree is 119898 (119898 ge 2) and in-degree is 0 (2) fortarget node its out-degree is 0 and in-degree is119898 (119898 ge 2) (3)for each intermediate node its out-degree is 1 and in-degreeis 1

Due to multiple serial trust chains we must considerall the impacts of them in calculation of trust value insteadof calculating them in an average method simply Here weproposed a local credit trust evaluation method for trustvalue calculation In this method only those trust chains inwhich the direct neighbor nodes keep high trust values withsource node would be considered total parallel trust valuecalculation

Let there be 119898 (119898 ge 2) serial trust chain between sourcenode 119873

119878and target node 119873

119864 119862119897denotes each serial trust

chain and 119879(119862119897) represents trust value of serial trust chain119862119897 For each 119862119897 119879(119873119878 119873

1

119868)119897 denotes the direct trust value of

atomic trust chain from 119873119878to its neighbor node 119873

1

119868in 119862119897

We set a threshold 120578 (120578 isin [0 1]) and define the serial trustchain which has condition119879(119873

119878 1198731

119868)119897le 120578 as ignorable chain

All ignorable trust chains are excluded in our local optimizedtrust evaluation Then the trust value of parallel trust chaincan be calculated as follows

119879 (119873119878 119873119864) =

1

119898

[

[

( sum

119879(119862119897)ge07

119879 (119862119897)11198981

)

+ ( sum

04lt119879(119862119897)lt07

119879 (119862119897)) + ( sum

119879(119862119897)le04

119879 (119862119897)

1198982

)]

]

(8)

In the above equation serial trust chains are divided intothree cases as follows (1) trust chains whose values are greaterthan or equal to 07 are considered as positive views towardtarget node and would be enlarged by exponential weightingas 1119898

1(here 119898

1denotes the number of positive views)

(2) trust chains whose values are between 04 and 07 areconsidered as neutral views toward target node and would betreated as their original values (3) trust chains whose valuesare less than or equal to 04 are considered as negative viewstoward target node and would be weakened by exponentialweighting as 119898

2(here 119898

2denotes the number of negative

views)

44 Combined Trust Chain Combined trust chain is com-posed of the above three kinds of trust chain That is therewould be a complex path composition from source node totarget node Here we address a scheme of combined trustchain value computation In this method several rules wouldbe followed in calculation as follows

(i) Local credit rule (LCR) only those trust chains inwhich the direct neighbor nodes keep high trustvalues with source node would be considered totalparallel trust value calculation That is the atomictrust chain whose trust value is lower than a thresholdcan be ignored in the combined trust chain Mean-while their successor trust chains are ignored sincethe paths are broken

(ii) Serial calculating rule (SCR) if there is a combinedtrust chain from119873119878rsquos indirect neighbor nodes to 119873119864it would be eliminated as a serial one recursively

(iii) Parallel calculating rule (PCR) if there are two ormore direct neighbor nodes of119873

119878 the combined trust

chain would be reconstructed as parallel trust chainfor119873119878with its neighbors recursively

An example of our scheme is shown in Figures 3(a)ndash3(d) and we can see how it works to calculate the combinedtrust chain In original combined trust chain there is acomplex combined path from source node to target nodethrough intermediate nodes of AndashM We use our proposedrules to reconstruct the combined trust chain as follows(1) in Figure 3(b) the atomic trust chains including sourcenode rarr B source node rarr C D rarr I and F rarr I areignored according to LCR (ignored chains are in dotted linesin Figure 3(b) and threshold 120578 is set as 03) Meanwhileatomic trust chains including B rarr G C rarr I and I rarr Mare ignored because their preorder trust chains are ignored(2) the part of combined trust chain E rarr target node isparallelized into two parallel trust chains by PCR E rarr G rarr

D rarr target node and E rarr K rarr target node (3) the part of Erarr target node is eliminated as a serial trust chain as E rarr Krarr target node by SCRTherefore the original combined trustchain can be reconstructed as a parallel trust chain which iscomposed of source node rarr A rarr G rarr K rarr target node(1198621) source node rarr E rarr K rarr target node (1198622) sourcenode rarr F rarr H rarr L rarr target node (119862

3) and source node

rarr D rarr J rarr M rarr target node (1198624)Then we can calculate

the combined trust chain as follows

119879 (1198621) =

1

4

times [1

4+ 09

3+ 07

2+ 08] = 075

119879 (1198622) =

1

3

times [09

3+ 088

2+ 08] = 077

119879 (1198623) =

1

4

times [07

4+ 1

3+ 09

2+ 1] = 076

119879 (1198624) =

1

4

times [08

4+ 09

3+ 1

2+ 07] = 071

119879 (119873119878 119873119864) =

1

4

times [075

13+ 077

13+ 076

13+ 071]

= 086

(9)

5 Method of Sentiment Modeling

51 Rules for Sentiment Modeling Sentiment is expressedin usersrsquo texts through evidences such as approving and


B

C

D

E

F

G

H

I

J

M

L

K

A

07

1

07

01

03

08

1

09

09

08

08

109

08

09

07

1

03

03 1

06

09

Source node

Intermediate nodeH

Target node

Atomic trust chain and its trust value 08

(a) Original combined trust chain

B

C

D

E

F

G

H

I

J

M

L

K

Ignorable trust chain

07

1

07

01

03

08

1

09

09

08

08

109

08

09

07

1

03

03

03 1

06

09

A

Source node

Intermediate nodeH

Target node


(b) Using LCR for ignoring low creditable neighbors

K

D

E

F

G

H

J

M

L

KG

Source node

Intermediate nodeH

Target node

A

(c) Using PCR for reconstructing trust chain

K

D

E

F

G

H

J

M

L

K

Series trust chain by using SCR

A

Source node

Intermediate nodeH

Target node

(d) Using SCR for eliminating parallel parts in trust chain

Figure 3 Example for combined trust chain calculation

forwarding with positive comments In this work sentimentis described formally by a sentiment ontology which providesformal and normalized knowledge and then enables thesentiment to be understood by system automatically Thesentiment ontology is built up manually because there is noexisting relevant ontology which is suitable for supportingthe following sentiment calculation rules All the classes and

their relationships are set manually according to knowledgeof sentiment category

Each dimension of sentiment 119904119890119894 = ⟨119905119910119901119890 119877119890119897119886119905119890119889V119886119897119906119890⟩ can find a corresponding ontology knowledgedescription through mapping function These dimensionsof sentiment are initial inputs for sentiment model Thenwe propose several rules to eliminate related dimensions


and ensure that the sentiment model is stable Assumethat the initial sentiment is 119878119864 119894119899 = 119904119890

1 1199041198902 Then

the dimensions can be normalized and calculated by thefollowing rules

(1) Class Merging Rule (CMR) Sentiment dimensions canbe merged if and only if there are ancestor-descendantrelationships or parent-children relationships between theirclasses in sentiment ontology

Here we first define an operator denoted ⊙ to cal-culate dimension changing based on TRP Let there betwo dimensions as 119904119890119894 = ⟨119905119910119901119890 119877119890119897119886119905119890119889 119897119890V119890119897⟩ and 119904119890119895 =

⟨119905119910119901119890 119877119890119897119886119905119890119889 119897119890V119890119897⟩ which satisfy CMR calculation and wedenote the new dimension as 119904119890119894 ⊙ 119904119890119895 = 119904119890119896 Then thecalculation based on CMR can be composed of the followingfour phases firstly we get the sentiment ontology mappingfunctions 119891 119904119890

119894119905119910119901119890 rarr 119878119874119888119897119886119904119904

119894and 119891 119904119890

119895119905119910119901119890 rarr

119878119874119888119897119886119904119904119895 secondly values of layer and level are obtained

according to ontology tree and then value of 119904119890119896is calculated

asV119886119897119906119890 (119904119890

119896) = V119886119897119906119890 (119904119890

119894) ⊙ V119886119897119906119890 (119904119890

119895)

=

(119897119890V119890119897 (119904119890119894) + 119897119886119910 (119888119897119886119904119904119894)) + (119897119890V119890119897 (119904119890119895) + 119897119886119910 (119888119897119886119904119904119895))

2

(10)

Then we can update the layer value of 119904119890119896

as119897119886119910119890119903(119891(119904119890

119896)) = lfloorV119886119897119906119890(119904119890

119896119905119910119901119890)rfloor and its level value as

119897119890V119890119897(119904119890119896) = V119886119897119906119890(119904119890119896) minus 119897119886119910119890119903(119891(119904119890119896119905119910119901119890)) further theclass of 119904119890119896 is determined according to the layer value inthe ontology tree and then the type of 119904119890119896 is 119904119890119896119905119910119901119890 =

119891minus1(119904119890119896119878119874119888119897119886119904119904) and finally we can get 119904119890

119896119877119890119897119886119905119890119889 =

119904119890119896119903119890119897119886119905119890119889119896 | 119904119890119896119903119890119897119886119905119890119889119896 = 119891

minus1(119891(119904119890119896119905119910119901119890)119877119890119897119886119905119894119900119899)

For example we can calculate the value of new sentimentdimension 1199041198905 which is generated by sentiment dimensionsof 1199041198901 and 1199041198904 if there is a CMR rule for them in Figure 2 as

V119886119897119906119890 (1199041198905)

=

(119897119890V119890119897 (1199041198901) + 119897119886119910 (119888119897119886119904119904

2)) + (119897119890V119890119897 (119904119890

4) + 119897119886119910 (119888119897119886119904119904

8))

2

=

(1 + 09) + (3 + 1)

2

= 295

(11)

Then we can get the layer value of 1199041198905as 119897119886119910119890119903(119891(119904119890

119896119905119910119901119890)) =

lfloor295rfloor = 2 and 119891(119904119890119896119905119910119901119890) = 119888119897119886119904119904

6according to

the ontology tree Meanwhile 119897119890V119890119897(1199041198905) = V119886119897119906119890(119904119890

5) minus

119897119886119910119890119903(119891(1199041198905119905119910119901119890)) = 095

Moreover let there be a set of sentiment dimensionswhich satisfies rule of CMR and they can bemerged into onedimension through CMR (119898119904119890) The merging calculation isas follows

119898119904119890 =

119899

⨀

119894=1

119904119890119894 (12)

where 119899 is the number of dimensions in the set

(2) Sentiment Migration Rule (SMR) Sentiment dimensionscan migrate into another one if and only if there are quali-tative changes under migration rules which are provided insentiment ontology

Sentiment migration rule is a kind of production ruleThat is SMR functions under legal reasoning rules whichare provided by ontology We also define an operator otimes tocalculate dimension changing based on SMR Let there betwo dimensions as 119904119890

119894= ⟨119905119910119901119890 119877119890119897119886119905119890119889 119897119890V119890119897⟩ and 119904119890

119895=

⟨119905119910119901119890 119877119890119897119886119905119890119889 119897119890V119890119897⟩ which satisfy SMR and we denote thenew dimension as 119904119890119894 otimes 119904119890

119895= 119904119890119896 Then the calculation of

SMR is as follows

119904119890119894otimes 119904119890119895= 119904119890119896

=

119904119890119896119905119910119901119890 = 119891

minus1[119903 (119891 (119904119890

119894) 119891 (119904119890

119895))]

119904119890119896119877119890119897119886119905119890119889 = 119891minus1

(119891 (119904119890119896119905119910119901119890) 119877119890119897119886119905119894119900119899)

119897119890V119890119897 (119904119890119896) =

119897119890V119890119897 (119904119890119894) + 119897119890V119890119897 (119904119890119895) minus 1

2

(13)

Here is an example of SMR which enables dimensions ofldquogladrdquo and ldquosatisfyingrdquo tomigrate into ldquohappinessrdquo as follows

(exist119891 (1199041198901119905119910119901119890) = 119892119897119886119889 and 119897119890V119890119897 (119904119890

1) ge 07)

and (exist119891 (1199041198902119905119910119901119890) = 119904119886119905119894119904119891119910119894119899119892 and 119897119890V119890119897 (119904119890

2) ge 08)

and (119903 (119892119897119886119889 119904119886119905119894119904119891119910119894119899119892) = ℎ119886119901119901119894119899119890119904119904)

997888rarr (exist (1199041198903119905119910119901119890) = ℎ119886119901119901119894119899119890119904119904 and 119897119890V119890119897 (119904119890

3)

=

119897119890V119890119897 (1199041198901) + 119897119890V119890119897 (119904119890

2) minus 1

2

)

(14)

(3) Time Merging Rule (TMR) Dimensions have theirdynamic change formats if and only if they are changed withthe time passing and satisfy CMR

TMR enables sentiment dimension to be evaluated withtime passing and reflects the dynamic feature of userrsquossentiment Here we propose a dynamic iterative estimatingmethod for describing the TMP based sentiment changingSuppose the original sentiment dimension is 1199041198900

119894and its next

value ismeasured to be 119888ℎ119886119899119892 1199041198901

119894after a next time sliceThen

we can update the dynamic estimating result by using CMRcalculation of these two sentiments that is

119904119890

1

119894= 119904119890

0

119894⊙ 119888ℎ119886119899119892 119904119890

1

119894 (15)

By analogy the dynamic estimating result of TMR can becalculated iteratively as

119904119890

119899

119894= 119904119890

119899minus1

119894⊙ 119888ℎ119886119899119892 119904119890

119899

119894 (16)

(4) Time Attenuation Rule (TAR) Level value of dimensionwould damp if and only if there is no new changes with thetime passing

Humanrsquos sentiment bent would be on the wane with thetime passing if there is not any new evidence whichmaintainsthis sentiment TAR is proposed to reflect such attenuationeffect In this study TAR only impacts the value of level fora sentiment dimension Let there be a sentiment dimension119904119890119894

= ⟨119905119910119901119890 119877119890119897119886119905119890119889 119897119890V119890119897⟩ and there is an attenuation


parameter 120573 the level value of dimension can be calculatedwith TAR as

119897119890V119890119897 (119904119890119894)119899= 119897119890V119890119897 (119904119890

119894)119899minus1

120573 (17)

where 119897119890V119890119897(119904119890119894)119899is the level value of 119904119890

119894after 119899 time slices

(119897119890V119890119897(119904119890119894)0is the original value here)

(5) Mutually Exclusive Rule (MER) Dimensions cannot becalculated if and only if there is no rule which can be usedfor them

MER reveals that the sentiment dimensions whichbelong to different branches in ontology tree are regarded ascompletely different ones unless they can be calculated underrules of CMR SMR and TPR

52 Sentiment Modeling Sentiment model describes all sen-timent dimensions in a stable state In this study sentimentmodeling method aims to normalize original sentimentdimensions through the above five rules for sentiment mod-eling We propose the brief modeling process here as follows

(i) All related sentiment dimensions which satisfy ruleof CMR are merged in original sentiment dimensionset

(ii) All sentiment dimensions which satisfy rule of SMRare calculated for modeling new dimensions

(iii) All sentiment dimensions which satisfy rules of TMRand TAR are remodeled at every time slice

6 Sentiment Delivering Estimate Scheme

61 Explicit Sentiment Delivering Estimate Scheme In mobilesocial network usersrsquo sentiments often are impacted by thosewho obtain usersrsquo trustworthiness For example user119860mightbe very angry towards a topic if his most trusted friend 119861

holds a very angry sentiment towards the same topic Suchsentiment delivering is shown in Figure 1(c) Therefore wecan estimate userrsquos explicit sentiment through hisher trustchains Here we address an explicit sentiment deliveringestimate scheme according to the trust chain

In our consideration peoplersquos explicit sentiment which isexpressed in text or other ways explicitly in past would beinfluenced by those who heshe considers trustworthy Andthen the explicit sentiment would be influenced and changedin some degrees while the user expresses it in next time(called explicit delivering sentiment) Our explicit sentimentdelivering estimate scheme aims to evaluate the userrsquos explicitsentiment in near future based on his past explicit sentimentand his trust chains Here we have the following descriptions

(i) Source node the user node who keeps explicit senti-ment model in mobile social network

(ii) Target node the user node who has expressed itsexplicit sentiment and would be influenced by sourcenode to form new explicit sentiment in near futurethrough their trust chain

(iii) Single explicit (implicit) sentiment one explicit(implicit) sentiment delivered from one source node

(iv) Integrated explicit (implicit) sentiment the explicit(implicit) sentiment which is composed of all singleexplicit (implicit) sentiments from all source nodes

In this study an iterative estimate evaluation methodis proposed for explicit sentiment delivering estimate Themethod is composed of two steps single explicit sentimentcalculation and integrated explicit sentiment evaluation Thetwo steps are as follows

(1) Single Delivering Explicit Sentiment Calculation In thisstep we calculate all single explicit sentiments from onesource node to the target node iteratively based on explicitsentiment model and their trust chains Assume that thereis a trust chain Ω(119873119878

119873119864) between two user nodes and the

past source explicit sentiment model of 119873119878and 119873

119864is 119878119864119873119878

and 119878119864119873119864

respectively Suppose that the trust chain value ofΩ(119873119864 119873119878) is 119879(119873

119878 119873119864) ge 120572 And then for each dimension

in single explicit delivering sentiment 119889119890119897119894V119890119903 119904119890119896(Ω) isin

119889119890119897119894V119890119903 119878119864(Ω) it is generated by 119878119864119873119878

119904119890119894 and 119878119864

119873119878

119904119890119894 and

is calculated as follows

(i) If there is CMR for 119878119864119873119878

119904119890119894and 119878119864

119873119864

119904119890119895 then the

single explicit delivering sentiment dimension can becalculated as

(119889119890119897119894V119890119903 119904119890119896 (Ω) = 119878119864

119873119878

119904119890119894⊙ 119878119864119873119864

119904119890119895)

and (V119886119897119906119890 (119889119890119897119894V119890119903 119904119890119896 (Ω)) = V119886119897119906119890 (119878119864

119873119878

119904119890119894)

⊙ V119886119897119906119890 (119878119864119873119864

119904119890119895) 119879 (119873119878 119873119864))

(18)

(ii) If there is SMR for 119878119864119873119878

119904119890119894 and 119878119864119873119864

119904119890119895 then thesingle explicit delivering sentiment dimension can becalculated as

119889119890119897119894V119890119903 119904119890119896 (Ω) = 119878119864

119873119878

119904119890119894otimes 119878119864119873119864

119904119890119895times Ω (119873

119864 119873119878)

=

119904119890119896119905119910119901119890 = 119891

minus1[119903 (119891 (119904119890

119894) 119891 (119904119890

119895))]

119904119890119896119877119890119897119886119905119890119889 = 119891

minus1(119891 (119904119890

119896119905119910119901119890) 119877119890119897119886119905119894119900119899)

119897119890V119890119897 (119904119890119896) =

119897119890V119890119897 (119904119890119894) + 119897119890V119890119897 (119904119890

119895) minus 1

2

119879 (119873119864 119873119878)

(19)

(iii) If there is no rule for 119878119864119873119864

119904119890119895 then the single explicit

delivering sentiment dimension can be calculated as

119889119890119897119894V119890119903 119904119890119896 (Ω) = 119878119864

119873119864

119904119890119895 (20)

(iv) If there is no rule for 119878119864119873119878



119889119890119897119894V119890119903 119904119890119896 (Ω)

=

119889119890119897119894V119890119903 119904119890119896 (Ω) 119905119910119901119890 = 119878119864

119873119878

119904119890119894119905119910119901119890

119889119890119897119894V119890119903 119904119890119896 (Ω) 119877119890119897119886119905119890119889 = 119878119864

119873119878

119904119890119894119877119890119897119886119905119890119889

119889119890119897119894V119890119903 119904119890119896 (Ω) 119897119890V119890119897 = 119878119864119873119878

119904119890119894119897119890V119890119897 times 119879 (119873119878 119873119864)

(21)


Then single explicit sentiment model from trust chainΩ(119873119864 119873119878) is described as

119889119890119897119894V119890119903 119878119864 (Ω)

= 119889119890119897119894V119890119903 1199041198901 (Ω) 119889119890119897119894V119890119903 119904119890

2 (Ω)

(22)

(2) Integrated Explicit Sentiment Evaluation In this step allsingle explicit sentiment models from different source nodesare integrated into one explicit sentiment model based onproposed five rules for sentiment calculation

Let there be a set of single explicit sentiment mod-els ⋃119889119890119897119894V119890119903 119878119864(Ω

119895) from different trust chains Then all

sentiment dimensions in 119889119890119897119894V119890119903 119878119864(Ω119895) can be extractedand then a new integrated explicit sentiment model can begenerated as

119889119890119897119894V119890119903 119878119864119873119864

= 119889119890119897119894V119890119903 119904119890119896(Ω119895) | 119889119890119897119894V119890119903 119904119890

119894(Ω119895)

isin 119889119890119897119894V119890119903 119878119864 (Ω119895) and 119889119890119897119894V119890119903 119878119864 (Ω

119895)

isin⋃119889119890119897119894V119890119903 119878119864 (Ω119895)

(23)

Then we use the following steps to reduce the number ofdimensions in 119889119890119897119894V119890119903 119878119864

119873119864

(i) All dimensions in 119889119890119897119894V119890119903 119878119864119873119864

are calculated basedon rules of CMR and SMR to reduce the number ofdimensions

(ii) The dimensions which satisfy the condition of119897119890V119890119897(119889119890119897119894V119890119903 119904119890

119894) le 120574 are regarded as noncritical

dimensions and are deleted from the model(iii) All dimensions which satisfy rules of TMR and TAR

are updated at every time slice(iv) The dimensions which satisfy the condition of

119897119890V119890119897(119889119890119897119894V119890119903 119904119890119894)119899 le 120574 are deleted as noncriticaldimensions after 119899 time slices

62 Implicit Sentiment Delivering Estimate Scheme Implicitsentiment is a kind of hidden emotion which reflects usersrsquopotential opinion In our view trust chains among usersdenote the probability of userrsquos inherent attitudes based ontheir trust relationships Here we address an estimatingevaluation for calculating the implicit sentiment according tothe trust chain

We consider that peoplersquos sentiment would be influencedby those who heshe trusts in For example a user mightapprove of a topic because his best friends keep positivefeelings toward the topic even though he does not show anyexplicit evidence Therefore the implicit sentiment can becalculated through the trust chains That is implicit senti-ment is transmitted through the chains Similar to explicitsentiment delivering estimate scheme we here propose animplicit sentiment delivering method which is composed oftwo steps single implicit sentiment delivering calculation andintegrated implicit delivering sentiment evaluation The twosteps are as follows

(1) Single Implicit Sentiment Delivering Calculation Assumethat there is a trust chain Ω(119873

119878 119873119864) between two user

nodes and the source explicit sentiment model of 119873119878is

119878119864119873119878

Suppose that the trust chain value of Ω(119873119864 119873119878) is

119879(119873119878 119873119864) ge 120572 And then for each dimension in single

implicit sentiment 119894119898119901119897119894119888119894119905 119904119890119894(Ω) isin 119894119898119901119897119894119888119894119905 119878119864(Ω) it is

generated by 119878119864119873119878

119904119890119894 which is delivered from 119873

119878 and is

calculated as

119894119898119901119897119894119888119894119905 119904119890119894 (Ω) 119905119910119901119890 = 119878119864

119873119878

119904119890119894119905119910119901119890

119894119898119901119897119894119888119894119905 119904119890119894 (Ω) 119877119890119897119886119905119890119889 = 119878119864

119873119878

119904119890119894119877119890119897119886119905119890119889

119894119898119901119897119894119888119894119905 119904119890119894 (Ω) 119897119890V119890119897 = 119878119864

119873119878

119904119890119894119897119890V119890119897 times 119879 (119873

119878 119873119864)

(24)

Then single implicit sentiment model from trust chainΩ(119873119864 119873119878) is described as

119894119898119901119897119894119888119894119905 119878119864 (Ω)

= 119894119898119901119897119894119888119894119905 1199041198901 (Ω) 119894119898119901119897119894119888119894119905 119904119890

2 (Ω)

(25)

(2) Integrated Implicit Delivering Sentiment Evaluation In thisstep all single implicit sentiment models from source nodesare integrated into one implicit sentiment model based onproposed five rules for sentiment calculation

Let there be a set of single implicit sentiment mod-els ⋃119894119898119901119897119894119888119894119905 119878119864(Ω


sentiment dimensions in 119894119898119901119897119894119888119894119905 119878119864(Ω119895) can be extractedand then a new integrated implicit sentiment model can begenerated as

119894119898119901119897119894119888119894119905 119878119864119873119864

= 119894119898119901119897119894119888119894119905 119904119890119894(Ω119895) | 119894119898119901119897119894119888119894119905 119904119890

119894(Ω119895)

isin 119894119898119901119897119894119888119894119905 119878119864 (Ω119895) and 119894119898119901119897119894119888119894119905 119878119864 (Ω

119895)

isin⋃ 119894119898119901119897119894119888119894119905 119878119864 (Ω119895)

(26)

Similar to explicit sentiment we use the following stepsto reduce the number of dimensions in 119894119898119901119897119894119888119894119905 119878119864119873

119864

(i) All dimensions in 119894119898119901119897119894119888119894119905 119878119864119873119864


(ii) The dimensions which satisfy the condition of119897119890V119890119897(119894119898119901119897119894119888119894119905 119904119890




119897119890V119890119897(119894119898119901119897119894119888119894119905 119904119890119894)119899

le 120574 are deleted as noncriticaldimensions after 119899 time slices

7 Experiment and Analysis

In this section we proposed examinations to explain the effi-ciency of our proposed method In our scenario of examina-tions the data comes from theTencentWeibowhich is a popu-larmobile social network platform inChinaWe collect infor-mation manually from httpsinacom microblog within a


Table 1 Characteristics of five communities in examination prototype

Community Social life Finance Sporting Entertainment Technology TotalNumber of IDs 913 792 1183 1054 854 3016Number of posts 3858 2693 4692 3013 3065 17321Number of comments 89076 86871 90562 49065 69651 385225

span of three months through the API interface which isprovided by httpsinacom Our data included about 3016IDs (some IDs located in two or more communities) andmore than 400000 records (including posts and comments)from October 2013 to February 2014 In the examinationprototype there is a one-way direct link from a user towardsanother one if heshe followed the user in httpsinacommicroblog All direct links (atomic trust chain) are generatedfrom initial data set and are fixed and invariable in prototypeAll these data are used to calculate the initial trust and formnetwork topology based on real-world source

Based on data we develop a prototype formonitoring andevaluating the effectiveness of our proposed method in thisstudy In our prototype nodes can be controlled by exper-imenters for examining requirements For initial settingthere are 5 communities that is social life finance sportingentertainment and technology We classify all IDs into 5communities according to their verified identifications (oneIDmight locate in two or more communities) Detailed char-acteristics of five communities in the prototype are shownin Table 1 There are four kinds of behaviors in the proto-type posting transmitting judging and acceptingrejectingMeanwhile atomic trust chains between IDs which havedirect interaction such as publishing judgments transfer-ring or following are set initially by calculating their pastinteraction according to the collected data In addition thenetwork topology of our prototype is generated according totrust chain relationships and average out-degree of a node is 7

In addition we construct sentiment ontology in ourexamination for sentiment delivering evaluation In thisontology we collected 107 sentiment classes for sentimentsemantic description There are 1091 relations and 923 rulesin ontology for sentiment modeling

71 Examination for Trust Chain Calculation The first exam-ination aims to testify the effectiveness of trust chainWe ver-ify the effectiveness of trust chain by adding about 800 addi-tional nodes which can be set as honest or malicious nodes inthe prototype All additional nodes are deployed randomlyAnd then we testify the trust chain calculationmethod underno malicious node and malicious nodes In the examinationwe define that a trust chain is accurate if the trust chain valuetoward an honest node is larger than 05 or the trust chainvalue toward a malicious node is lower than 04

711 Accuracy of Four Types of Trust Chain In this examina-tion we compare the calculationmethods of the four types oftrust chain that is atom trust chain (ATC) serial trust chain(STC) parallel trust chain (PTC) and combined trust chain(CTC) and record their accuracies We test 10000 timesof interaction under no malicious node and 15 malicious

nodesThe results are shown in Figures 4(a) and 4(b)We cansee that the average accuracy of atom trust chain obtains thebest performance in the examination Meanwhile combinedtrust chain and parallel trust chain have close accuracieswhich reach up to 904 861 896 and 838 underno malicious node and 15 malicious nodes We think thatthe reasons are as follows (1) the long distance of path ispunished in BF and our methods (2) confidence value isimpacted by the reputation weight in our proposed methodso that the confidence would be attenuated if there were poorreputation nodes in the path and (3) parallel path confidenceis a comprehensive evaluation result based on all single pathsin the parallel path which leads to better effects than singlepath confidence

In addition we verify the impacts of parameter 120578 fortrust chain calculation We test the different values of 120578 forparallel trust chain and combined trust chain respectivelyIn Figure 4(c) we can see that the accuracy is low while thevalue is 120578 too low or too high On the basis of test validationa threshold around 06ndash07 is often a reasonable compromiseIn our consideration the reasons are as follows (1) someuntrustworthy neighbors are taken into account if the value ofthreshold 120578 is set too low which would decrease the accuracyof parallel and combined trust chain and (2) reasonableneighbors and their following trust chains are neglected ifthe value of threshold 120578 is set too high which leads to theaccuracy of parallel and combined trust chain decrease

712 PerformanceComparison of ProposedTrust ChainCalcu-lation In this examination we first revealed the effectivenessof our proposed trust chain calculationmethod by comparingwith other existing methods Here we selected nodes whichlocated in our prototype at random and then calculatedtheir trust chain values through our proposedmethods (TC)For comparison we set the other two methods trust withBellman-Ford algorithm (BF) [15] and average method oftrust aggregation (AT) [6] The results are shown in Figures5(a) and 5(b) After 10000 times of comparison averageaccuracy of our proposed parallel confidence method isbetter than other methods under no malicious node and 15malicious nodes environments respectively In our analysisthe reasons are as follows (1) the long distance of path ispunished in BF and our methods (2) trust value is impactedby the reputation weight in our proposed method so that thetrust value would be attenuated if there were poor reputationnodes in the path and (3) trust chain value is a comprehensiveevaluation result based on all composition paths in the chainwhich leads to better effects than other methods

Moreover we verified the impacts of the numbers ofpaths communities and depths In Figure 5(c) averageaccuracies of trust calculation are decreased with the number


Accu

racy

of t

rust

chai

n ca

lcul

atio

n (

)

75

80

85

90

95

4000 6000 8000 100002000Number of calculations

ATCCTCPTCSTC

(a)

Accu

racy

of t

rust

chai

n ca

lcul

atio

n (

)

70

75

80

85

90

95


ATCCTCPTCSTC

(b)

Accu

racy

of t

rust

chai

n ca

lcul

atio

n (

)

40

50

60

70

80

90

100

Combined trust chainParallel trust chain

20 806040 1000Value of threshold 120578

(c)

Figure 4 Effects of trust chain calculation

of paths increasing in BF and AT while the accuracy ofproposed trust chain kept increasing That is because theshortest path is admitted for trust computation in BF andall paths are treated equally in AT which results in theignorance of other significant paths in BF and mistakenlyadoption of malicious paths or nodes in AT In our methodall trustable paths (impacted by parameter 120578) were computedcomprehensively and all nodes are impacted by their rep-utation based on related rules Figure 5(d) shows that withthe numbers of paths (trust chain) and depths increasingthe accuracies were decreasing In BF method accuracy wasnearly stable because long paths are deemed untrustworthyIn our method trust chain values are attenuated with thedepth increasing according to depth attenuation impacts in(7) From this point a larger value of depth in trust chain alsoimplies that the trust chain would be an untrustworthy oneFigure 5(e) shows the accuracies of trust calculation under

different numbers of communities which trust chain pathspassed throughWe can see that the accuracy of our proposedtrust chain was higher than the other two methods with thenumber of communities increasing

72 Performance Evaluation of Sentiment Modeling In thisexamination we aim to verify the effectiveness and feasibilityof sentiment modeling method based on proposed rulesin this study The initial setting of sentiment ontology isshown as the beginning of this section Firstly we testifythe feasibility of our proposed rules for sentiment modelingWe tracked the sentiments which were expressed by userstowards specific topics and recorded the average accuraciesof sentiment modeling under rules of CMR SMR TARand TMR We selected 300 users at random and modeledtheir sentiments in their posts and comments In Figure 6(a)results show that the accuracies of our proposed four rules


70

80

90

100Av

erag

e acc

urac

y (

)

TCBFAT

2000 4000Number of interactions

6000 8000 10000

(a) No malicious node

40

60

50

70

80

90

100

Aver

age a

ccur

acy

()

TC BFAT


6000 8000 10000

(b) 15malicious nodes

50

60

70

80

90

100

Aver

age a

ccur

acy

()

2 3 4 5 6 7 81Number of paths

TCBF

AT

(c) 15malicious nodes

4 6 82 10Number of depths

50

60

70

80

90

100Av

erag

e acc

urac

y (

)

TCBF

AT

(d) 15malicious nodes

50

60

70

80

90

100

Aver

age a

ccur

acy

()

2 3 4 51Number of communities

TCBF

AT

(e) 15malicious nodes

Figure 5 Performance comparison of proposed trust chain calculation


60

70

80

90

100

Aver

age a

ccur

acy

()

TARTMRSMRCMR

50403010 20Number of days

(a)

0

20

40

60

80

100

Aver

age a

ccur

acy

()

20 40 60 80 1000Value of factor 120573

(b)

0

20

40

60

80

100

Accu

racy

()

ESMLDAMESVMKM(c) Social life

ESMLDAMESVMKM0

20

40

60

80

100

Accu

racy

()

(d) Finance

ESMLDAMESVMKM0

20

40

60

80

100

Accu

racy

()

(e) SportsESMLDAMESVMKM

0

20

40

60

80

100

Accu

racy

()

(f) Entertainment

Figure 6 Continued


ESMLDAMESVMKM0

20

40

60

80

100

Accu

racy

()

(g) Technology

Figure 6 Performance of explicit sentiment modeling

are acceptable after 50 days of tracking In addition we verifythe impact of factor 120573 in rule TAR Figure 6(b) shows that theaccuracies of sentiment evaluation are low while the value offactor 120573 is too low or too high We can see that a thresholdaround 07ndash09 is often a reasonable compromise and theaccuracy is around 86ndash89

Further we reveal the performance of proposed sen-timent modeling method by comparing existing methodsIn our examination we set three methods for comparisonas sentiment analysis based on keyword matching (KM)sentiment classification based on support vector machines(SVM) max entropy method (ME) LDA model (LDA) andour proposed sentimentmodeling (ESM)methodThe resultsare shown in Figures 6(c)ndash6(g) which reveal the accuraciesof sentiment analysis in our initial setting communities Wecan see that the LDA method had the best performanceswhile the accuracies of our proposed sentiment modelingwere almost equal to LDAmethodThat means the proposedmethod is feasible and effective in sentiment analysis

73 Performance Evaluation of Explicit Sentiment DeliveringEstimate Scheme This examination reveals the effectivenessof explicit sentiment delivering estimate scheme We selectedabout 1000 topics in our dataset for evaluating explicitsentiment delivering In the test we recorded the accuracy ofsentiment estimate based on past explicit sentiment recordsFor verifying the effects on sentiment estimate we set datafrom September 2013 to December 2013 as past data forpriori probability training preparation Then we testified theaccuracies of results with the microblog data from January2014 to February 2014 For comparison we set four groupsas maximum likelihood estimate method (MLE) the prioriestimate method (PE) Bayesian method (BM) and ourproposed explicit sentiment delivering method (ESD) Theresults are shown in Figures 7(a) and 7(b) In Figure 7(a) werecorded the accuracies of sentiment delivering for extremesentiment including anger blame guilt excitement andhappiness in different groups while accuracies in Figure 7(b)were recorded for normal sentiment As a result of our

analysis the performances of probability-based estimatemethods rely on the training dataset in past Therefore withthe number of estimate times increasing the accuracies wereincreasing in groups of MLE PE and BM Compared withthe three methods our method mainly relies on the effectsof trust chain evaluation and sentiment ontology Then wecan see that our method got the best performances whichwere around 74 and 79 respectively We also notice thatthe accuracies were higher than other methods from thebeginning In our consideration the reasons are as follows(1) the sentiment ontology is constructed in advance so thatthe knowledge and rules of explicit sentiment delivering areaccurate at the beginning (2) the trust chains among users areestablished in advance and the sentiment delivering is basedon accurate trust chains (3) trust chains can be renewedaccording to results of sentiment delivering to ensure the highaccuracy of trust chain

Moreover we verify the impacts of parameter 120572 forexplicit sentiment delivering We test the different values of 120572for accuracy of sentiment delivering estimate In Figure 7(c)we can see that the average accuracy is lower while the valueof 120572 is lower than 06 In addition we notice that the accuracyis decreased while the value of 120572 is larger than 08 Fromour perspective that is because many valuable trust chainswhich may impact the explicit sentiment delivering resultwere ignored in such case On the basis of test validation avalue around 06ndash08 is often a reasonable compromise Inour other examinations its default value is set as 07

74 Performance Evaluation of Implicit Sentiment DeliveringEstimate Scheme In this examination we aim to testify theeffectiveness and feasibility of proposed implicit sentimentdelivering estimate scheme The examination setting is sim-ilar to examinations in Section 74 We evaluated implicitsentiments of users which were never expressed in theirpast posts and comments explicitly if they had transferringfollowing and similar behaviors toward specific topics in thedataset from September 2013 to December 2013 And thenwe verified the usersrsquo sentiments towards the topics if they


40

50

60

70

80

90

100Av

erag

e acc

urac

y (

)

1000 1500 2000 2500 3000500Number of sentiment estimates

ESDBMPEMLE

(a) Extreme sentiment delivering

40

50

60

70

80

90

100

Aver

age a

ccur

acy

()


ESDBMPEMLE

(b) Normal sentiment delivering

6040 1000 20 800

20

40

60

80

100

Aver

age a

ccur

acy

()

Value of threshold 120572

(c)

Figure 7 Examinations of explicit sentiment delivering estimate

expressed their sentiments explicitly in the last two monthsFor comparison we set two groups as user relationshipbased implicit sentiment estimate scheme (URM) in whichdelivering factor is set as 1 and our proposed trust chain basedmethod (TCM) The results are shown in Figures 8(a) and8(b) We recorded the accuracies of the above two methodsin collected five communities that is social life (SL) finance(FI) sports (SP) entertainment (EN) and technology (TE)Figure 8(a) shows that the accuracies of extreme sentimentsdelivering estimate was around 547 while the accuracies ofnormal sentiments delivering estimate was around 458 Inour consideration extreme sentimentwould havemore prob-abilities to be delivered than normal sentiment through usersrsquotrustworthiness in trust chain which resulted in that theaccuracy of extreme sentiment delivering estimate was higherthan accuracy of normal sentiment delivering estimate

Likewise we verify the impacts of parameter 120574 for implicitsentiment delivering estimate We test the different values of120574 for accuracy In Figure 8(c) we can see that the average

accuracy is lower while the value of 120574 is lower than 02 orlarger than 04 And also we can get that a value around02ndash04 is often a reasonable compromise In our otherexaminations its default value is set as 03

8 Conclusion

Sentiment analysis is essential for mobile social networkbecause of its convenient communication pattern and thenature of gathering mass of public opinions Sentiments insuch relationship oriented platform are delivered throughusersrsquo trustworthy relationships and then are impactedmutu-ally As a result sentiment evaluation in mobile socialnetwork should be considered from a delivering view Mostexisting researches indicate that mining explicit sentimentis feasible and effective However few of them take thesentiment delivering estimate into consideration In thisstudy sentiment which includes explicit sentiment andimplicit sentiment is analyzed based on its delivering nature


URMTCM

TEENSPFISL0

10

20

30

40

50

60

70Av

erag

e acc

urac

y (

)

(a) Extreme sentiment

URMTCM

TEENSPFISL0

10

20

30

40

50

60

70

Aver

age a

ccur

acy

()

(b) Normal sentiment

6040 1000 20 800

20

40

60

80

100

Aver

age a

ccur

acy

()


(c)

Figure 8 Examinations of implicit sentiment delivering estimate

and the usersrsquo trustworthy relationships In our work trustchain which is the model of describing and evaluatingtrustworthiness among users is defined and its calculationmethod is addressed according to its path composition asatomic path serial path parallel path and combined pathSentiment in this work is described in a formal semanticmethod based on sentiment ontology Then the methodof explicit sentiment modeling is proposed through a setof sentiment modeling rules Further delivering estimatescheme for explicit sentiment and implicit sentiment isproposed based on trust chain and sentiment modeling Infuture studies we aim to identify a mechanism of groupsentiment delivering mining for user community or group inmobile social network

Conflict of Interests

The authors declare that there is no conflict of interestsregarding the publication of this paper

Acknowledgments

This work is funded by National Basic Research Program ofChina (2014CB340404) National Natural Science Founda-tion of China (61572326 61103069 71171148 and 61272268)Innovation Program of Shanghai Municipal Education Com-mission (13YZ052) the Programof Shanghai NormalUniver-sity (DCL201302) and the Shanghai Rising-Star Program (no15QA1403900)

References

[1] P Groenewegen and C Moser ldquoOnline communities chal-lenges and opportunities for mobile social network researchrdquoResearch in the Sociology of Organizations vol 40 pp 463ndash4772014

[2] D Boyd S Golder and G Lotan ldquoTweet tweet retweetconversational aspects of retweeting on twitterrdquo in Proceedingsof the 43rd Annual Hawaii International Conference on SystemSciences (HICSS rsquo10) pp 1ndash10 IEEE Honolulu Hawaii USAJanuary 2010


[3] H He ldquoSentiment analysis of Sina Weibo based on semanticsentiment space modelrdquo in Proceedings of the 20th InternationalConference on Management Science and Engineering (ICMSErsquo13) pp 206ndash211 IEEE Harbin China July 2013

[4] B Pang and L Lee ldquoOpinion mining and sentiment analysisrdquoFoundations and Trends in Information Retrieval vol 2 no 1-2pp 1ndash135 2008

[5] MMMostafa ldquoMore than words social networksrsquo text miningfor consumer brand sentimentsrdquo Expert Systems with Applica-tions vol 40 no 10 pp 4241ndash4251 2013

[6] A Joslashsang R Ismail and C Boyd ldquoA survey of trust and rep-utation systems for online service provisionrdquo Decision SupportSystems vol 43 no 2 pp 618ndash644 2007

[7] E W K See-To and K K W Ho ldquoValue co-creation andpurchase intention in mobile social network sites the role ofelectronic Word-of-Mouth and trustmdasha theoretical analysisrdquoComputers in Human Behavior vol 31 pp 182ndash189 2014

[8] J Wu and F Chiclana ldquoA social network analysis trustndashconsensus based approach to group decision-making problemswith interval-valued fuzzy reciprocal preference relationsrdquoKnowledge-Based Systems vol 59 pp 97ndash107 2014

[9] G Wang and X-L Gui ldquoSelecting and trust computing fortransaction nodes in online social networksrdquo Jisuanji Xuebaovol 36 no 2 pp 368ndash383 2013

[10] D Li Q Lv X Xie et al ldquoInterest-based real-time contentrecommendation in online social communitiesrdquo Knowledge-Based Systems vol 28 pp 1ndash12 2012

[11] Z Peiyun C Enhong and L Bo ldquoWeb services trust computa-tion based onmobile social network dynamic feedbackrdquoPatternRecognition and Artificial Intelligence vol 26 no 4 pp 337ndash3432013

[12] X-Q Qiao C Yang X-F Li and J-L Chen ldquoA trust calculatingalgorithm based on social networking service usersrsquo contextrdquoJisuanji Xuebao vol 34 no 12 pp 2403ndash2413 2011

[13] F Javier Ortega J A Troyano F L Cruz C G Vallejo and FEnrıquez ldquoPropagation of trust and distrust for the detectionof trolls in a social networkrdquoComputer Networks vol 56 no 12pp 2884ndash2895 2012

[14] B Qureshi G Min and D Kouvatsos ldquoTrusted informationexchange in peer-to-peer mobile social networksrdquo ConcurrencyComputation Practice and Experience vol 24 no 17 pp 2055ndash2068 2012

[15] H Zhao and X Li ldquoVectorTrust trust vector aggregationscheme for trust management in peer-to-peer networksrdquo inProceedings of the 18th International Conference on ComputerCommunications and Networks (ICCCN rsquo09) pp 1ndash6 IEEE SanFrancisco Calif USA August 2009

[16] J Golbeck ldquoTrust and nuanced profile similarity in online socialnetworksrdquoACMTransactions on theWeb vol 3 no 4 article 122009

[17] N Li and D D Wu ldquoUsing text mining and sentiment analysisfor online forums hotspot detection and forecastrdquo DecisionSupport Systems vol 48 no 2 pp 354ndash368 2010

[18] S Tan X Cheng Y Wang and H Xu ldquoAdapting naive bayesto domain adaptation for sentiment analysisrdquo in Advances inInformation Retrieval vol 5478 of Lecture Notes in ComputerScience pp 337ndash349 Springer Berlin Germany 2009

[19] J Boyd-Graber and P Resnik ldquoHolistic sentiment analysisacross languages multilingual supervised latent Dirichlet allo-cationrdquo in Proceedings of the Conference on Empirical Methodsin Natural Language Processing (EMNLP rsquo10) pp 45ndash55 Asso-ciation for Computational Linguistics October 2010

[20] M Thelwall and K Buckley ldquoTopic-based sentiment analysisfor the social web the role of mood and issue-related wordsrdquoJournal of the American Society for Information Science andTechnology vol 64 no 8 pp 1608ndash1617 2013

[21] E Cambria B Schuller B Liu H Wang and C HavasildquoKnowledge-based approaches to concept-level sentiment anal-ysisrdquo IEEE Intelligent Systems vol 28 no 2 pp 12ndash14 2013

[22] F Greaves D Ramirez-Cano C Millett A Darzi and LDonaldson ldquoUse of sentiment analysis for capturing patientexperience from free-text comments posted onlinerdquo Journal ofMedical Internet Research vol 15 no 11 article e239 2013

[23] E Kontopoulos C Berberidis T Dergiades and N BassiliadesldquoOntology-based sentiment analysis of twitter postsrdquo ExpertSystems with Applications vol 40 no 10 pp 4065ndash4074 2013

[24] T Nasukawa and J Yi ldquoSentiment analysis capturing favorabil-ity using natural language processingrdquo in Proceedings of the 2ndInternational Conference on Knowledge Capture (K-CAP rsquo03)pp 70ndash77 ACM Sanibel Island Fla USA October 2003

[25] P Goncalves M Araujo F Benevenuto andM Cha ldquoCompar-ing and combining sentiment analysis methodsrdquo in Proceedingsof the 1st ACMConference onOnline Social Networks (COSN rsquo13)pp 27ndash37 ACM Boston Mass USA October 2013

[26] Q Su X Xu H Guo et al ldquoHidden sentiment associationin Chinese web opinion miningrdquo in Proceedings of the 17thInternational Conference on World Wide Web (WWW rsquo08) pp959ndash968 ACM Beijing China April 2008

[27] A Athar and S Teufel ldquoDetection of implicit citations forsentiment detectionrdquo in Proceedings of the Workshop on Detect-ing Structure in Scholarly Discourse (DSSD rsquo12) pp 18ndash26Association for Computational Linguistics July 2012

[28] A Balahur J M Hermida and AMontoyo ldquoDetecting implicitexpressions of sentiment in text based on commonsense knowl-edgerdquo in Proceedings of the 2nd Workshop on ComputationalApproaches to Subjectivity and Sentiment Analysis (WASSA rsquo11)Association forComputational Linguistics PortlandOreUSAJune 2011

[29] A Balahur J M Hermida and AMontoyo ldquoDetecting implicitexpressions of emotion in text a comparative analysisrdquoDecisionSupport Systems vol 53 no 4 pp 742ndash753 2012

[30] X Zhou X Tao J Yong and Z Yang ldquoSentiment analysison tweets for social eventsrdquo in Proceedings of the IEEE 17thInternational Conference on Computer Supported CooperativeWork in Design (CSCWD rsquo13) pp 557ndash562 Whistler CanadaJune 2013

[31] T Dergiades C Milas and T Panagiotidis ldquoTweets Googletrends and sovereign spreads in the GIIPSrdquo Oxford EconomicPapers vol 67 no 2 pp 406ndash432 2015

[32] MThelwall DWilkinson and S Uppal ldquoDatamining emotionin social network communication gender differences in MyS-pacerdquo Journal of the American Society for Information Scienceand Technology vol 61 no 1 pp 190ndash199 2010

[33] T H Soliman M A Elmasry A Hedar and M M DossldquoSentiment analysis of Arabic slang comments on facebookrdquoInternational Journal of Computers amp Technology vol 12 no 5pp 3470ndash3478 2014

[34] A Severyn A Moschitti O Uryupina B Plank and K Filip-pova ldquoMulti-lingual opinion mining on YouTuberdquo InformationProcessing amp Management 2015

[35] C Baecchi T Uricchio M Bertini and A Del Bimbo ldquoAmultimodal feature learning approach for sentiment analysis ofsocial networkmultimediardquoMultimedia Tools and Applications2015


[36] C Clavel and Z Callejas ldquoSentiment analysis from opinionmining to human-agent interactionrdquo IEEE Transactions onAffective Computing 2016

Submit your manuscripts athttpwwwhindawicom

Computer Games Technology

International Journal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014


Distributed Sensor Networks


Advances in

FuzzySystems

Hindawi Publishing Corporationhttpwwwhindawicom

Volume 2014


ReconfigurableComputing

Hindawi Publishing Corporation httpwwwhindawicom Volume 2014


Applied Computational Intelligence and Soft Computing

thinspAdvancesthinspinthinsp

Artificial Intelligence

HindawithinspPublishingthinspCorporationhttpwwwhindawicom Volumethinsp2014

Advances inSoftware EngineeringHindawi Publishing Corporationhttpwwwhindawicom Volume 2014


Electrical and Computer Engineering

Journal of

Journal of

Computer Networks and Communications


Hindawi Publishing Corporation

httpwwwhindawicom Volume 2014

Advances in

Multimedia


Biomedical Imaging


ArtificialNeural Systems

Advances in


RoboticsJournal of



Computational Intelligence and Neuroscience

Industrial EngineeringJournal of


Modelling amp Simulation in EngineeringHindawi Publishing Corporation httpwwwhindawicom Volume 2014

The Scientific World JournalHindawi Publishing Corporation httpwwwhindawicom Volume 2014


Human-ComputerInteraction

Advances in

Computer EngineeringAdvances in



drives sentiment delivering is the trustworthiness amongusersThus by employing trust as the basis for our sentimentdelivering evaluation ourmain contributions in this work areas follows (1) proposing an overview of sentiment deliveringevaluation scheme is addressed and its related definitionsincluding trust chain sentiment semantics and sentimentontology are proposed (2) presenting a formal trust chainmodel which links users and measures the trustworthinessis proposed (3) presenting a sentiment modeling methodis presented based on a set of rules which enable sentimentsemantic to be calculated through definitions in sentimentontology and (4) proposing a sentiment delivering estimatescheme is therefore addressed which is composed of explicitand implicit sentiment estimate based on trust chain andsentiment modeling

The rest of the paper is organized as follows Section 2discusses related work of our study Section 3 presents anoverview of scheme and its related definitions Section 4addresses the trust chainmodel and its trust value calculationmethod Sections 5 and 6 present the modeling method anddelivering estimate schemes of explicit sentiment and implicitsentiment respectively Empirical results and discussion aregiven in Section 7 Finally Section 8 concludes the paper

2 Related Work

21 Trust Computation Trust as an inherent willingness ofhuman beings shows the emotional and rational confidencein between people It is derived from judging trustworthinessby evaluating various facts which can lead to either the con-fidence or distrust Over the past few years many works havefocused on computing trusted paths in network environmentsuch as P2P and mobile social networks (SNS) [6ndash10] Intrust evaluation there are two core kinds of trust as directtrust and indirect trust Direct trust is used for reflectingthe trustworthiness between direct connected users whileindirect trust is used widely in long path connected usersthrough intermediate users [6] Typically many historicaldata based on direct interactions such as behaviors com-ments or other various evidences are used to calculate thedirect trust degree among users [6] In most cases the directtrust reflects the subjective trustworthy feelings towards thetargets through the historical direct interactions See-To andHo [7] propose a method to evaluate the influence of truston oral comment in mobile social network Wu and Chiclana[8] present an approach to computing trust that users have allagreed on in mobile social network Wang and Gui [9] selecttransaction nodes in mobile social network and computetrust between them Based on real-time content Li et al [10]develop a community information recommendation systemwhich takes advantage of user experience Peiyun et al [11]propose an algorithm for computing direct trust value andfurthermore develop a Web service trust model based onmobile social network dynamic feedback Qiao et al [12]propose a context-based trust computation method wherethe trust generation principle in the psychology is employedIn addition the indirect trust is to evaluate trustworthinessthrough a deliverable perspective of trust concerning theindirect connected relationships among users [13ndash15] From

such point direct trust evaluation is also the foundationof indirect trust evaluation In most cases the calculationof direct and indirect trust is discussed based on a graphmodel direct trust can be seen as a direct connected edgebetween vertices while the indirect trust is seen as a longpath composed of direct trust in graph model Therefore thepath among indirect connected users is an essential factor forindirect trust evaluation Javier Ortega et al [13] propose amethod to compute a ranking of the users in a mobile socialnetwork and propagate both positive andnegative opinions ofthe users Then the opinions from each user about others caninfluence their global trust score including direct and indi-rect trust views Qureshi et al [14] propose a decentralizedframework and the related algorithms for trusted informationexchange and social interaction among users based on thedynamicity aware graph relabeling systemTheBellman-Fordalgorithm computes trust based on direct witness interactiontrust judgments [15] It generates a trust graph on the basis ofthe trust link between two peers who have direct interactionEach peer can submit or renew their trust judgments of othersbased on new direct interactions Further the trust betweenpeers is constantly updated by compounding old and newtrust judgments In addition the algorithm admits the mosttrustable path for trust computation it deems a long pathto be untrustworthy Golbeck proposed TidalTrust that getstrust in mobile social networks using numeric trust values[16] It utilized the shortest path based on the breadth-firstsearch Further TidalTrust can be used to retrieve accurateinformation from the highest trust adjacent nodes

However most of these above methods focus on eitherdirect trust or reputation computation and do not take intoaccount the chain relationship of trust among users andits complex path topology Different from traditional trustmodules there are the following considerations in this study(1) since the sentiments among users are considered to bedelivered through the user relationships the proposed trustmodel in this study therefore takes chain relationships amongusers into account to reflect the indirect confidencewhich canbe used to measure the probability of sentiment deliveringthrough users by their mutual trustworthiness (2) for directtrust calculation that is atomic trust chain in this workwe also use the weighted average method to calculate themwhich is similar to other traditional works However weintroduce a new factor of community which widely existsin mobile social network for weights evaluation That is thedirect trust calculation is impacted by the facts whether theusers are in same community or not and (3) with respect toindirect trust the proposed trust model calculates it throughuser indirect relationships The relationship compositionamong users that is the path topological information amongusers in social network graph is complex and variousUnfortunately few attentions have been paid to the complexpath composition for indirect trust That is there are manydifferent accessible paths between users and each accessiblepath manifests a delivering confidence Then the proposedtrust model aims to calculate the trust chain with taking allpaths confidence into account comprehensively in mobilesocial network which is also one of the main motivationswhich we work in this study




















UserTrust chain


Topic


A B C




A D E




1 1199041198902 in which 119904119890

119894denotes one



119894=



1199041198901 119889119890119897119894119892ℎ119905 = ⟨119905119910119901119890 ℎ119886119901119901119894119899119890119904119904 119877119890119897119886119905119890119889

119895119900119910119891119906119897 | ⟨SMRTAR⟩ 119890119888119904119905119886119904119910 | ⟨SMRTMR⟩

119897119890V119890119897 09⟩

1199041198902 119908119900119903119903119910 = ⟨119905119910119901119890 119891119890119886119903 119877119890119897119886119905119890119889

119904119886119889119899119890119904119904 | ⟨SMRTAR⟩ ℎ119886119905119890 | ⟨TMR⟩ 119897119890V119890119897 03⟩

(1)










119878119874 = ⟨119862119897119886119904119904 119877119900119900119905 ℎ119886119901119901119894119899119890119904119904 119904119886119889119899119890119904119904 119886119899119892119890119903

119877119890119897119886119905119894119900119899 119901119886119903119890119899119905-119888ℎ119894119897119889 ⟨119877119900119900119905 ℎ119886119901119901119894119899119890119904119904⟩

119901119886119903119890119899119905-119888ℎ119894119897119889 ⟨119877119900119900119905 119904119886119889119899119890119904119904⟩

119901119886119903119890119899119905-119888ℎ119894119897119889 ⟨119877119900119900119905 119886119899119892119890119903⟩ 119877119906119897119890 CMR SMR

TMRTARMER⟩

(2)






119894119905119910119901119890 rarr 119878119874119888119897119886119904119904

119895


119894 119888119897119886119904119904

119895) = 119888119897119886119904119904


classes 119888119897119886119904119904119894and 119888119897119886119904119904



119894 119888119897119886119904119904

119895)


SMR 119903 (119905119890119903119903119891119894119890119889 119908119900119903119903119910) = 119889119894119904119905119906119903119887119890119889 (3)




as119897119886119910119890119903 (1198881198971198861199041199042) = 1

119897119886119910119890119903 (1198881198971198861199041199045) = 2

119897119886119910119890119903 (1198881198971198861199041199048) = 3

119891 1199041198901119905119910119901119890 997888rarr 1198781198741198881198971198861199041199042

119891 1199041198902119905119910119901119890 997888rarr 1198781198741198881198971198861199041199045

119891 1199041198904119905119910119901119890 997888rarr 1198781198741198881198971198861199041199048

(4)






1198732and


2isin 1198662)) 119891119894(1198731 1198732)


Level(se4) = 1

Level(se3) = 08

Level(se2) = 08

1Level(se ) = 09

se4

se3

se2

se1


Sentiment dimension




Class 2Class 1

Root




1


1 1198732) can be


119879 (1198731 1198732) =

0 119899 = 0

sum

119899

119894=1(119891119894(1198731 1198732) times 119896119894)

119899

119899 ge 1

(5)





119896119894 =





(6)




119864 and




119879119862(119873119878 119873119864) =

1

119889119890119901119905ℎ (119873119864)

[119879 (119873119878 119873

1

119868)

119889119890119901119905ℎ(119873119864)

+

119889119890119901119905ℎ(119873119864)minus1

sum

119896=2

119879 (119873

119896

119868 119873

119896+1

119868)

119889119890119901119905ℎ(119873119864)minus119896+1

+ 119879 (119873

119889119890119901119905ℎ(119873119864)minus1

119868 119873119864)]

(7)













1



1

119868in 119862119897


119878 1198731



119879 (119873119878 119873119864) =

1

119898

[

[

( sum

119879(119862119897)ge07

119879 (119862119897)11198981

)

+ ( sum

04lt119879(119862119897)lt07

119879 (119862119897)) + ( sum

119879(119862119897)le04

119879 (119862119897)

1198982

)]

]

(8)


1(here 119898



2(here 119898


views)









3) and source node



119879 (1198621) =

1

4

times [1

4+ 09

3+ 07

2+ 08] = 075

119879 (1198622) =

1

3

times [09

3+ 088

2+ 08] = 077

119879 (1198623) =

1

4

times [07

4+ 1

3+ 09

2+ 1] = 076

119879 (1198624) =

1

4

times [08

4+ 09

3+ 1

2+ 07] = 071

119879 (119873119878 119873119864) =

1

4

times [075

13+ 077

13+ 076

13+ 071]

= 086

(9)




B

C

D

E

F

G

H

I

J

M

L

K

A

07

1

07

01

03

08

1

09

09

08

08

109

08

09

07

1

03

03 1

06

09

Source node

Intermediate nodeH

Target node



B

C

D

E

F

G

H

I

J

M

L

K


07

1

07

01

03

08

1

09

09

08

08

109

08

09

07

1

03

03

03 1

06

09

A

Source node

Intermediate nodeH

Target node



K

D

E

F

G

H

J

M

L

KG

Source node

Intermediate nodeH

Target node

A


K

D

E

F

G

H

J

M

L

K


A

Source node

Intermediate nodeH

Target node








1 1199041198902 Then





119894119905119910119901119890 rarr 119878119874119888119897119886119904119904

119894and 119891 119904119890

119895119905119910119901119890 rarr



asV119886119897119906119890 (119904119890

119896) = V119886119897119906119890 (119904119890

119894) ⊙ V119886119897119906119890 (119904119890

119895)

=

(119897119890V119890119897 (119904119890119894) + 119897119886119910 (119888119897119886119904119904119894)) + (119897119890V119890119897 (119904119890119895) + 119897119886119910 (119888119897119886119904119904119895))

2

(10)


as119897119886119910119890119903(119891(119904119890

119896)) = lfloorV119886119897119906119890(119904119890




119896119877119890119897119886119905119890119889 =

119904119890119896119903119890119897119886119905119890119889119896 | 119904119890119896119903119890119897119886119905119890119889119896 = 119891

minus1(119891(119904119890119896119905119910119901119890)119877119890119897119886119905119894119900119899)


V119886119897119906119890 (1199041198905)

=

(119897119890V119890119897 (1199041198901) + 119897119886119910 (119888119897119886119904119904

2)) + (119897119890V119890119897 (119904119890

4) + 119897119886119910 (119888119897119886119904119904

8))

2

=

(1 + 09) + (3 + 1)

2

= 295

(11)


119896119905119910119901119890)) =


6according to


5) minus

119897119886119910119890119903(119891(1199041198905119905119910119901119890)) = 095


119898119904119890 =

119899

⨀

119894=1

119904119890119894 (12)




119894= ⟨119905119910119901119890 119877119890119897119886119905119890119889 119897119890V119890119897⟩ and 119904119890

119895=



SMR is as follows

119904119890119894otimes 119904119890119895= 119904119890119896

=

119904119890119896119905119910119901119890 = 119891

minus1[119903 (119891 (119904119890

119894) 119891 (119904119890

119895))]

119904119890119896119877119890119897119886119905119890119889 = 119891minus1

(119891 (119904119890119896119905119910119901119890) 119877119890119897119886119905119894119900119899)

119897119890V119890119897 (119904119890119896) =

119897119890V119890119897 (119904119890119894) + 119897119890V119890119897 (119904119890119895) minus 1

2

(13)


(exist119891 (1199041198901119905119910119901119890) = 119892119897119886119889 and 119897119890V119890119897 (119904119890

1) ge 07)

and (exist119891 (1199041198902119905119910119901119890) = 119904119886119905119894119904119891119910119894119899119892 and 119897119890V119890119897 (119904119890

2) ge 08)

and (119903 (119892119897119886119889 119904119886119905119894119904119891119910119894119899119892) = ℎ119886119901119901119894119899119890119904119904)

997888rarr (exist (1199041198903119905119910119901119890) = ℎ119886119901119901119894119899119890119904119904 and 119897119890V119890119897 (119904119890

3)

=

119897119890V119890119897 (1199041198901) + 119897119890V119890119897 (119904119890

2) minus 1

2

)

(14)



119894and its next




119904119890

1

119894= 119904119890

0

119894⊙ 119888ℎ119886119899119892 119904119890

1

119894 (15)


119904119890

119899

119894= 119904119890

119899minus1

119894⊙ 119888ℎ119886119899119892 119904119890

119899

119894 (16)






119897119890V119890119897 (119904119890119894)119899= 119897119890V119890119897 (119904119890

119894)119899minus1

120573 (17)






















119864is 119878119864119873119878

and 119878119864119873119864





119904119890119894 and 119878119864

119873119878

119904119890119894 and



119904119890119894and 119878119864

119873119864

119904119890119895 then the


(119889119890119897119894V119890119903 119904119890119896 (Ω) = 119878119864

119873119878

119904119890119894⊙ 119878119864119873119864

119904119890119895)

and (V119886119897119906119890 (119889119890119897119894V119890119903 119904119890119896 (Ω)) = V119886119897119906119890 (119878119864

119873119878

119904119890119894)

⊙ V119886119897119906119890 (119878119864119873119864

119904119890119895) 119879 (119873119878 119873119864))

(18)


119904119890119894 and 119878119864119873119864


119889119890119897119894V119890119903 119904119890119896 (Ω) = 119878119864

119873119878

119904119890119894otimes 119878119864119873119864

119904119890119895times Ω (119873

119864 119873119878)

=

119904119890119896119905119910119901119890 = 119891

minus1[119903 (119891 (119904119890

119894) 119891 (119904119890

119895))]

119904119890119896119877119890119897119886119905119890119889 = 119891

minus1(119891 (119904119890

119896119905119910119901119890) 119877119890119897119886119905119894119900119899)

119897119890V119890119897 (119904119890119896) =

119897119890V119890119897 (119904119890119894) + 119897119890V119890119897 (119904119890

119895) minus 1

2

119879 (119873119864 119873119878)

(19)




119889119890119897119894V119890119903 119904119890119896 (Ω) = 119878119864

119873119864

119904119890119895 (20)




119889119890119897119894V119890119903 119904119890119896 (Ω)

=

119889119890119897119894V119890119903 119904119890119896 (Ω) 119905119910119901119890 = 119878119864

119873119878

119904119890119894119905119910119901119890

119889119890119897119894V119890119903 119904119890119896 (Ω) 119877119890119897119886119905119890119889 = 119878119864

119873119878

119904119890119894119877119890119897119886119905119890119889

119889119890119897119894V119890119903 119904119890119896 (Ω) 119897119890V119890119897 = 119878119864119873119878

119904119890119894119897119890V119890119897 times 119879 (119873119878 119873119864)

(21)



119889119890119897119894V119890119903 119878119864 (Ω)

= 119889119890119897119894V119890119903 1199041198901 (Ω) 119889119890119897119894V119890119903 119904119890

2 (Ω)

(22)





119889119890119897119894V119890119903 119878119864119873119864

= 119889119890119897119894V119890119903 119904119890119896(Ω119895) | 119889119890119897119894V119890119903 119904119890

119894(Ω119895)

isin 119889119890119897119894V119890119903 119878119864 (Ω119895) and 119889119890119897119894V119890119903 119878119864 (Ω

119895)

isin⋃119889119890119897119894V119890119903 119878119864 (Ω119895)

(23)


119873119864











119878 119873119864) between two user


119878119864119873119878




generated by 119878119864119873119878


119878 and is

calculated as

119894119898119901119897119894119888119894119905 119904119890119894 (Ω) 119905119910119901119890 = 119878119864

119873119878

119904119890119894119905119910119901119890

119894119898119901119897119894119888119894119905 119904119890119894 (Ω) 119877119890119897119886119905119890119889 = 119878119864

119873119878

119904119890119894119877119890119897119886119905119890119889

119894119898119901119897119894119888119894119905 119904119890119894 (Ω) 119897119890V119890119897 = 119878119864

119873119878

119904119890119894119897119890V119890119897 times 119879 (119873

119878 119873119864)

(24)


119894119898119901119897119894119888119894119905 119878119864 (Ω)

= 119894119898119901119897119894119888119894119905 1199041198901 (Ω) 119894119898119901119897119894119888119894119905 119904119890

2 (Ω)

(25)





119894119898119901119897119894119888119894119905 119878119864119873119864

= 119894119898119901119897119894119888119894119905 119904119890119894(Ω119895) | 119894119898119901119897119894119888119894119905 119904119890

119894(Ω119895)

isin 119894119898119901119897119894119888119894119905 119878119864 (Ω119895) and 119894119898119901119897119894119888119894119905 119878119864 (Ω

119895)

isin⋃ 119894119898119901119897119894119888119894119905 119878119864 (Ω119895)

(26)


119864







119897119890V119890119897(119894119898119901119897119894119888119894119905 119904119890119894)119899

















Accu

racy

of t

rust

chai

n ca

lcul

atio

n (

)

75

80

85

90

95


ATCCTCPTCSTC

(a)

Accu

racy

of t

rust

chai

n ca

lcul

atio

n (

)

70

75

80

85

90

95


ATCCTCPTCSTC

(b)

Accu

racy

of t

rust

chai

n ca

lcul

atio

n (

)

40

50

60

70

80

90

100



(c)






70

80

90

100Av

erag

e acc

urac

y (

)

TCBFAT


6000 8000 10000


40

60

50

70

80

90

100

Aver

age a

ccur

acy

()

TC BFAT


6000 8000 10000


50

60

70

80

90

100

Aver

age a

ccur

acy

()


TCBF

AT



50

60

70

80

90

100Av

erag

e acc

urac

y (

)

TCBF

AT


50

60

70

80

90

100

Aver

age a

ccur

acy

()


TCBF

AT




60

70

80

90

100

Aver

age a

ccur

acy

()

TARTMRSMRCMR


(a)

0

20

40

60

80

100

Aver

age a

ccur

acy

()

20 40 60 80 1000Value of factor 120573

(b)

0

20

40

60

80

100

Accu

racy

()


ESMLDAMESVMKM0

20

40

60

80

100

Accu

racy

()

(d) Finance

ESMLDAMESVMKM0

20

40

60

80

100

Accu

racy

()


0

20

40

60

80

100

Accu

racy

()

(f) Entertainment

Figure 6 Continued


ESMLDAMESVMKM0

20

40

60

80

100

Accu

racy

()

(g) Technology









40

50

60

70

80

90

100Av

erag

e acc

urac

y (

)


ESDBMPEMLE


40

50

60

70

80

90

100

Aver

age a

ccur

acy

()


ESDBMPEMLE


6040 1000 20 800

20

40

60

80

100

Aver

age a

ccur

acy

()


(c)





8 Conclusion



URMTCM

TEENSPFISL0

10

20

30

40

50

60

70Av

erag

e acc

urac

y (

)


URMTCM

TEENSPFISL0

10

20

30

40

50

60

70

Aver

age a

ccur

acy

()


6040 1000 20 800

20

40

60

80

100

Aver

age a

ccur

acy

()


(c)





Acknowledgments


References














































Advances in

FuzzySystems


Volume 2014












Journal of

Journal of





Advances in

Multimedia


Biomedical Imaging



Advances in


RoboticsJournal of










Advances in






















UserTrust chain


Topic


A B C




A D E




1 1199041198902 in which 119904119890

119894denotes one



119894=



1199041198901 119889119890119897119894119892ℎ119905 = ⟨119905119910119901119890 ℎ119886119901119901119894119899119890119904119904 119877119890119897119886119905119890119889

119895119900119910119891119906119897 | ⟨SMRTAR⟩ 119890119888119904119905119886119904119910 | ⟨SMRTMR⟩

119897119890V119890119897 09⟩

1199041198902 119908119900119903119903119910 = ⟨119905119910119901119890 119891119890119886119903 119877119890119897119886119905119890119889

119904119886119889119899119890119904119904 | ⟨SMRTAR⟩ ℎ119886119905119890 | ⟨TMR⟩ 119897119890V119890119897 03⟩

(1)










119878119874 = ⟨119862119897119886119904119904 119877119900119900119905 ℎ119886119901119901119894119899119890119904119904 119904119886119889119899119890119904119904 119886119899119892119890119903

119877119890119897119886119905119894119900119899 119901119886119903119890119899119905-119888ℎ119894119897119889 ⟨119877119900119900119905 ℎ119886119901119901119894119899119890119904119904⟩

119901119886119903119890119899119905-119888ℎ119894119897119889 ⟨119877119900119900119905 119904119886119889119899119890119904119904⟩

119901119886119903119890119899119905-119888ℎ119894119897119889 ⟨119877119900119900119905 119886119899119892119890119903⟩ 119877119906119897119890 CMR SMR

TMRTARMER⟩

(2)






119894119905119910119901119890 rarr 119878119874119888119897119886119904119904

119895


119894 119888119897119886119904119904

119895) = 119888119897119886119904119904


classes 119888119897119886119904119904119894and 119888119897119886119904119904



119894 119888119897119886119904119904

119895)


SMR 119903 (119905119890119903119903119891119894119890119889 119908119900119903119903119910) = 119889119894119904119905119906119903119887119890119889 (3)




as119897119886119910119890119903 (1198881198971198861199041199042) = 1

119897119886119910119890119903 (1198881198971198861199041199045) = 2

119897119886119910119890119903 (1198881198971198861199041199048) = 3

119891 1199041198901119905119910119901119890 997888rarr 1198781198741198881198971198861199041199042

119891 1199041198902119905119910119901119890 997888rarr 1198781198741198881198971198861199041199045

119891 1199041198904119905119910119901119890 997888rarr 1198781198741198881198971198861199041199048

(4)






1198732and


2isin 1198662)) 119891119894(1198731 1198732)


Level(se4) = 1

Level(se3) = 08

Level(se2) = 08

1Level(se ) = 09

se4

se3

se2

se1


Sentiment dimension




Class 2Class 1

Root




1


1 1198732) can be


119879 (1198731 1198732) =

0 119899 = 0

sum

119899

119894=1(119891119894(1198731 1198732) times 119896119894)

119899

119899 ge 1

(5)





119896119894 =





(6)




119864 and




119879119862(119873119878 119873119864) =

1

119889119890119901119905ℎ (119873119864)

[119879 (119873119878 119873

1

119868)

119889119890119901119905ℎ(119873119864)

+

119889119890119901119905ℎ(119873119864)minus1

sum

119896=2

119879 (119873

119896

119868 119873

119896+1

119868)

119889119890119901119905ℎ(119873119864)minus119896+1

+ 119879 (119873

119889119890119901119905ℎ(119873119864)minus1

119868 119873119864)]

(7)













1



1

119868in 119862119897


119878 1198731



119879 (119873119878 119873119864) =

1

119898

[

[

( sum

119879(119862119897)ge07

119879 (119862119897)11198981

)

+ ( sum

04lt119879(119862119897)lt07

119879 (119862119897)) + ( sum

119879(119862119897)le04

119879 (119862119897)

1198982

)]

]

(8)


1(here 119898



2(here 119898


views)









3) and source node



119879 (1198621) =

1

4

times [1

4+ 09

3+ 07

2+ 08] = 075

119879 (1198622) =

1

3

times [09

3+ 088

2+ 08] = 077

119879 (1198623) =

1

4

times [07

4+ 1

3+ 09

2+ 1] = 076

119879 (1198624) =

1

4

times [08

4+ 09

3+ 1

2+ 07] = 071

119879 (119873119878 119873119864) =

1

4

times [075

13+ 077

13+ 076

13+ 071]

= 086

(9)




B

C

D

E

F

G

H

I

J

M

L

K

A

07

1

07

01

03

08

1

09

09

08

08

109

08

09

07

1

03

03 1

06

09

Source node

Intermediate nodeH

Target node



B

C

D

E

F

G

H

I

J

M

L

K


07

1

07

01

03

08

1

09

09

08

08

109

08

09

07

1

03

03

03 1

06

09

A

Source node

Intermediate nodeH

Target node



K

D

E

F

G

H

J

M

L

KG

Source node

Intermediate nodeH

Target node

A


K

D

E

F

G

H

J

M

L

K


A

Source node

Intermediate nodeH

Target node








1 1199041198902 Then





119894119905119910119901119890 rarr 119878119874119888119897119886119904119904

119894and 119891 119904119890

119895119905119910119901119890 rarr



asV119886119897119906119890 (119904119890

119896) = V119886119897119906119890 (119904119890

119894) ⊙ V119886119897119906119890 (119904119890

119895)

=

(119897119890V119890119897 (119904119890119894) + 119897119886119910 (119888119897119886119904119904119894)) + (119897119890V119890119897 (119904119890119895) + 119897119886119910 (119888119897119886119904119904119895))

2

(10)


as119897119886119910119890119903(119891(119904119890

119896)) = lfloorV119886119897119906119890(119904119890




119896119877119890119897119886119905119890119889 =

119904119890119896119903119890119897119886119905119890119889119896 | 119904119890119896119903119890119897119886119905119890119889119896 = 119891

minus1(119891(119904119890119896119905119910119901119890)119877119890119897119886119905119894119900119899)


V119886119897119906119890 (1199041198905)

=

(119897119890V119890119897 (1199041198901) + 119897119886119910 (119888119897119886119904119904

2)) + (119897119890V119890119897 (119904119890

4) + 119897119886119910 (119888119897119886119904119904

8))

2

=

(1 + 09) + (3 + 1)

2

= 295

(11)


119896119905119910119901119890)) =


6according to


5) minus

119897119886119910119890119903(119891(1199041198905119905119910119901119890)) = 095


119898119904119890 =

119899

⨀

119894=1

119904119890119894 (12)




119894= ⟨119905119910119901119890 119877119890119897119886119905119890119889 119897119890V119890119897⟩ and 119904119890

119895=



SMR is as follows

119904119890119894otimes 119904119890119895= 119904119890119896

=

119904119890119896119905119910119901119890 = 119891

minus1[119903 (119891 (119904119890

119894) 119891 (119904119890

119895))]

119904119890119896119877119890119897119886119905119890119889 = 119891minus1

(119891 (119904119890119896119905119910119901119890) 119877119890119897119886119905119894119900119899)

119897119890V119890119897 (119904119890119896) =

119897119890V119890119897 (119904119890119894) + 119897119890V119890119897 (119904119890119895) minus 1

2

(13)


(exist119891 (1199041198901119905119910119901119890) = 119892119897119886119889 and 119897119890V119890119897 (119904119890

1) ge 07)

and (exist119891 (1199041198902119905119910119901119890) = 119904119886119905119894119904119891119910119894119899119892 and 119897119890V119890119897 (119904119890

2) ge 08)

and (119903 (119892119897119886119889 119904119886119905119894119904119891119910119894119899119892) = ℎ119886119901119901119894119899119890119904119904)

997888rarr (exist (1199041198903119905119910119901119890) = ℎ119886119901119901119894119899119890119904119904 and 119897119890V119890119897 (119904119890

3)

=

119897119890V119890119897 (1199041198901) + 119897119890V119890119897 (119904119890

2) minus 1

2

)

(14)



119894and its next




119904119890

1

119894= 119904119890

0

119894⊙ 119888ℎ119886119899119892 119904119890

1

119894 (15)


119904119890

119899

119894= 119904119890

119899minus1

119894⊙ 119888ℎ119886119899119892 119904119890

119899

119894 (16)






119897119890V119890119897 (119904119890119894)119899= 119897119890V119890119897 (119904119890

119894)119899minus1

120573 (17)






















119864is 119878119864119873119878

and 119878119864119873119864





119904119890119894 and 119878119864

119873119878

119904119890119894 and



119904119890119894and 119878119864

119873119864

119904119890119895 then the


(119889119890119897119894V119890119903 119904119890119896 (Ω) = 119878119864

119873119878

119904119890119894⊙ 119878119864119873119864

119904119890119895)

and (V119886119897119906119890 (119889119890119897119894V119890119903 119904119890119896 (Ω)) = V119886119897119906119890 (119878119864

119873119878

119904119890119894)

⊙ V119886119897119906119890 (119878119864119873119864

119904119890119895) 119879 (119873119878 119873119864))

(18)


119904119890119894 and 119878119864119873119864


119889119890119897119894V119890119903 119904119890119896 (Ω) = 119878119864

119873119878

119904119890119894otimes 119878119864119873119864

119904119890119895times Ω (119873

119864 119873119878)

=

119904119890119896119905119910119901119890 = 119891

minus1[119903 (119891 (119904119890

119894) 119891 (119904119890

119895))]

119904119890119896119877119890119897119886119905119890119889 = 119891

minus1(119891 (119904119890

119896119905119910119901119890) 119877119890119897119886119905119894119900119899)

119897119890V119890119897 (119904119890119896) =

119897119890V119890119897 (119904119890119894) + 119897119890V119890119897 (119904119890

119895) minus 1

2

119879 (119873119864 119873119878)

(19)




119889119890119897119894V119890119903 119904119890119896 (Ω) = 119878119864

119873119864

119904119890119895 (20)




119889119890119897119894V119890119903 119904119890119896 (Ω)

=

119889119890119897119894V119890119903 119904119890119896 (Ω) 119905119910119901119890 = 119878119864

119873119878

119904119890119894119905119910119901119890

119889119890119897119894V119890119903 119904119890119896 (Ω) 119877119890119897119886119905119890119889 = 119878119864

119873119878

119904119890119894119877119890119897119886119905119890119889

119889119890119897119894V119890119903 119904119890119896 (Ω) 119897119890V119890119897 = 119878119864119873119878

119904119890119894119897119890V119890119897 times 119879 (119873119878 119873119864)

(21)



119889119890119897119894V119890119903 119878119864 (Ω)

= 119889119890119897119894V119890119903 1199041198901 (Ω) 119889119890119897119894V119890119903 119904119890

2 (Ω)

(22)





119889119890119897119894V119890119903 119878119864119873119864

= 119889119890119897119894V119890119903 119904119890119896(Ω119895) | 119889119890119897119894V119890119903 119904119890

119894(Ω119895)

isin 119889119890119897119894V119890119903 119878119864 (Ω119895) and 119889119890119897119894V119890119903 119878119864 (Ω

119895)

isin⋃119889119890119897119894V119890119903 119878119864 (Ω119895)

(23)


119873119864











119878 119873119864) between two user


119878119864119873119878




generated by 119878119864119873119878


119878 and is

calculated as

119894119898119901119897119894119888119894119905 119904119890119894 (Ω) 119905119910119901119890 = 119878119864

119873119878

119904119890119894119905119910119901119890

119894119898119901119897119894119888119894119905 119904119890119894 (Ω) 119877119890119897119886119905119890119889 = 119878119864

119873119878

119904119890119894119877119890119897119886119905119890119889

119894119898119901119897119894119888119894119905 119904119890119894 (Ω) 119897119890V119890119897 = 119878119864

119873119878

119904119890119894119897119890V119890119897 times 119879 (119873

119878 119873119864)

(24)


119894119898119901119897119894119888119894119905 119878119864 (Ω)

= 119894119898119901119897119894119888119894119905 1199041198901 (Ω) 119894119898119901119897119894119888119894119905 119904119890

2 (Ω)

(25)





119894119898119901119897119894119888119894119905 119878119864119873119864

= 119894119898119901119897119894119888119894119905 119904119890119894(Ω119895) | 119894119898119901119897119894119888119894119905 119904119890

119894(Ω119895)

isin 119894119898119901119897119894119888119894119905 119878119864 (Ω119895) and 119894119898119901119897119894119888119894119905 119878119864 (Ω

119895)

isin⋃ 119894119898119901119897119894119888119894119905 119878119864 (Ω119895)

(26)


119864







119897119890V119890119897(119894119898119901119897119894119888119894119905 119904119890119894)119899

















Accu

racy

of t

rust

chai

n ca

lcul

atio

n (

)

75

80

85

90

95


ATCCTCPTCSTC

(a)

Accu

racy

of t

rust

chai

n ca

lcul

atio

n (

)

70

75

80

85

90

95


ATCCTCPTCSTC

(b)

Accu

racy

of t

rust

chai

n ca

lcul

atio

n (

)

40

50

60

70

80

90

100



(c)






70

80

90

100Av

erag

e acc

urac

y (

)

TCBFAT


6000 8000 10000


40

60

50

70

80

90

100

Aver

age a

ccur

acy

()

TC BFAT


6000 8000 10000


50

60

70

80

90

100

Aver

age a

ccur

acy

()


TCBF

AT



50

60

70

80

90

100Av

erag

e acc

urac

y (

)

TCBF

AT


50

60

70

80

90

100

Aver

age a

ccur

acy

()


TCBF

AT




60

70

80

90

100

Aver

age a

ccur

acy

()

TARTMRSMRCMR


(a)

0

20

40

60

80

100

Aver

age a

ccur

acy

()

20 40 60 80 1000Value of factor 120573

(b)

0

20

40

60

80

100

Accu

racy

()


ESMLDAMESVMKM0

20

40

60

80

100

Accu

racy

()

(d) Finance

ESMLDAMESVMKM0

20

40

60

80

100

Accu

racy

()


0

20

40

60

80

100

Accu

racy

()

(f) Entertainment

Figure 6 Continued


ESMLDAMESVMKM0

20

40

60

80

100

Accu

racy

()

(g) Technology









40

50

60

70

80

90

100Av

erag

e acc

urac

y (

)


ESDBMPEMLE


40

50

60

70

80

90

100

Aver

age a

ccur

acy

()


ESDBMPEMLE


6040 1000 20 800

20

40

60

80

100

Aver

age a

ccur

acy

()


(c)





8 Conclusion



URMTCM

TEENSPFISL0

10

20

30

40

50

60

70Av

erag

e acc

urac

y (

)


URMTCM

TEENSPFISL0

10

20

30

40

50

60

70

Aver

age a

ccur

acy

()


6040 1000 20 800

20

40

60

80

100

Aver

age a

ccur

acy

()


(c)





Acknowledgments


References














































Advances in

FuzzySystems


Volume 2014












Journal of

Journal of





Advances in

Multimedia


Biomedical Imaging



Advances in


RoboticsJournal of










Advances in









UserTrust chain


Topic


A B C




A D E




1 1199041198902 in which 119904119890

119894denotes one



119894=



1199041198901 119889119890119897119894119892ℎ119905 = ⟨119905119910119901119890 ℎ119886119901119901119894119899119890119904119904 119877119890119897119886119905119890119889

119895119900119910119891119906119897 | ⟨SMRTAR⟩ 119890119888119904119905119886119904119910 | ⟨SMRTMR⟩

119897119890V119890119897 09⟩

1199041198902 119908119900119903119903119910 = ⟨119905119910119901119890 119891119890119886119903 119877119890119897119886119905119890119889

119904119886119889119899119890119904119904 | ⟨SMRTAR⟩ ℎ119886119905119890 | ⟨TMR⟩ 119897119890V119890119897 03⟩

(1)










119878119874 = ⟨119862119897119886119904119904 119877119900119900119905 ℎ119886119901119901119894119899119890119904119904 119904119886119889119899119890119904119904 119886119899119892119890119903

119877119890119897119886119905119894119900119899 119901119886119903119890119899119905-119888ℎ119894119897119889 ⟨119877119900119900119905 ℎ119886119901119901119894119899119890119904119904⟩

119901119886119903119890119899119905-119888ℎ119894119897119889 ⟨119877119900119900119905 119904119886119889119899119890119904119904⟩

119901119886119903119890119899119905-119888ℎ119894119897119889 ⟨119877119900119900119905 119886119899119892119890119903⟩ 119877119906119897119890 CMR SMR

TMRTARMER⟩

(2)






119894119905119910119901119890 rarr 119878119874119888119897119886119904119904

119895


119894 119888119897119886119904119904

119895) = 119888119897119886119904119904


classes 119888119897119886119904119904119894and 119888119897119886119904119904



119894 119888119897119886119904119904

119895)


SMR 119903 (119905119890119903119903119891119894119890119889 119908119900119903119903119910) = 119889119894119904119905119906119903119887119890119889 (3)




as119897119886119910119890119903 (1198881198971198861199041199042) = 1

119897119886119910119890119903 (1198881198971198861199041199045) = 2

119897119886119910119890119903 (1198881198971198861199041199048) = 3

119891 1199041198901119905119910119901119890 997888rarr 1198781198741198881198971198861199041199042

119891 1199041198902119905119910119901119890 997888rarr 1198781198741198881198971198861199041199045

119891 1199041198904119905119910119901119890 997888rarr 1198781198741198881198971198861199041199048

(4)






1198732and


2isin 1198662)) 119891119894(1198731 1198732)


Level(se4) = 1

Level(se3) = 08

Level(se2) = 08

1Level(se ) = 09

se4

se3

se2

se1


Sentiment dimension




Class 2Class 1

Root




1


1 1198732) can be


119879 (1198731 1198732) =

0 119899 = 0

sum

119899

119894=1(119891119894(1198731 1198732) times 119896119894)

119899

119899 ge 1

(5)





119896119894 =





(6)




119864 and




119879119862(119873119878 119873119864) =

1

119889119890119901119905ℎ (119873119864)

[119879 (119873119878 119873

1

119868)

119889119890119901119905ℎ(119873119864)

+

119889119890119901119905ℎ(119873119864)minus1

sum

119896=2

119879 (119873

119896

119868 119873

119896+1

119868)

119889119890119901119905ℎ(119873119864)minus119896+1

+ 119879 (119873

119889119890119901119905ℎ(119873119864)minus1

119868 119873119864)]

(7)













1



1

119868in 119862119897


119878 1198731



119879 (119873119878 119873119864) =

1

119898

[

[

( sum

119879(119862119897)ge07

119879 (119862119897)11198981

)

+ ( sum

04lt119879(119862119897)lt07

119879 (119862119897)) + ( sum

119879(119862119897)le04

119879 (119862119897)

1198982

)]

]

(8)


1(here 119898



2(here 119898


views)









3) and source node



119879 (1198621) =

1

4

times [1

4+ 09

3+ 07

2+ 08] = 075

119879 (1198622) =

1

3

times [09

3+ 088

2+ 08] = 077

119879 (1198623) =

1

4

times [07

4+ 1

3+ 09

2+ 1] = 076

119879 (1198624) =

1

4

times [08

4+ 09

3+ 1

2+ 07] = 071

119879 (119873119878 119873119864) =

1

4

times [075

13+ 077

13+ 076

13+ 071]

= 086

(9)




B

C

D

E

F

G

H

I

J

M

L

K

A

07

1

07

01

03

08

1

09

09

08

08

109

08

09

07

1

03

03 1

06

09

Source node

Intermediate nodeH

Target node



B

C

D

E

F

G

H

I

J

M

L

K


07

1

07

01

03

08

1

09

09

08

08

109

08

09

07

1

03

03

03 1

06

09

A

Source node

Intermediate nodeH

Target node



K

D

E

F

G

H

J

M

L

KG

Source node

Intermediate nodeH

Target node

A


K

D

E

F

G

H

J

M

L

K


A

Source node

Intermediate nodeH

Target node








1 1199041198902 Then





119894119905119910119901119890 rarr 119878119874119888119897119886119904119904

119894and 119891 119904119890

119895119905119910119901119890 rarr



asV119886119897119906119890 (119904119890

119896) = V119886119897119906119890 (119904119890

119894) ⊙ V119886119897119906119890 (119904119890

119895)

=

(119897119890V119890119897 (119904119890119894) + 119897119886119910 (119888119897119886119904119904119894)) + (119897119890V119890119897 (119904119890119895) + 119897119886119910 (119888119897119886119904119904119895))

2

(10)


as119897119886119910119890119903(119891(119904119890

119896)) = lfloorV119886119897119906119890(119904119890




119896119877119890119897119886119905119890119889 =

119904119890119896119903119890119897119886119905119890119889119896 | 119904119890119896119903119890119897119886119905119890119889119896 = 119891

minus1(119891(119904119890119896119905119910119901119890)119877119890119897119886119905119894119900119899)


V119886119897119906119890 (1199041198905)

=

(119897119890V119890119897 (1199041198901) + 119897119886119910 (119888119897119886119904119904

2)) + (119897119890V119890119897 (119904119890

4) + 119897119886119910 (119888119897119886119904119904

8))

2

=

(1 + 09) + (3 + 1)

2

= 295

(11)


119896119905119910119901119890)) =


6according to


5) minus

119897119886119910119890119903(119891(1199041198905119905119910119901119890)) = 095


119898119904119890 =

119899

⨀

119894=1

119904119890119894 (12)




119894= ⟨119905119910119901119890 119877119890119897119886119905119890119889 119897119890V119890119897⟩ and 119904119890

119895=



SMR is as follows

119904119890119894otimes 119904119890119895= 119904119890119896

=

119904119890119896119905119910119901119890 = 119891

minus1[119903 (119891 (119904119890

119894) 119891 (119904119890

119895))]

119904119890119896119877119890119897119886119905119890119889 = 119891minus1

(119891 (119904119890119896119905119910119901119890) 119877119890119897119886119905119894119900119899)

119897119890V119890119897 (119904119890119896) =

119897119890V119890119897 (119904119890119894) + 119897119890V119890119897 (119904119890119895) minus 1

2

(13)


(exist119891 (1199041198901119905119910119901119890) = 119892119897119886119889 and 119897119890V119890119897 (119904119890

1) ge 07)

and (exist119891 (1199041198902119905119910119901119890) = 119904119886119905119894119904119891119910119894119899119892 and 119897119890V119890119897 (119904119890

2) ge 08)

and (119903 (119892119897119886119889 119904119886119905119894119904119891119910119894119899119892) = ℎ119886119901119901119894119899119890119904119904)

997888rarr (exist (1199041198903119905119910119901119890) = ℎ119886119901119901119894119899119890119904119904 and 119897119890V119890119897 (119904119890

3)

=

119897119890V119890119897 (1199041198901) + 119897119890V119890119897 (119904119890

2) minus 1

2

)

(14)



119894and its next




119904119890

1

119894= 119904119890

0

119894⊙ 119888ℎ119886119899119892 119904119890

1

119894 (15)


119904119890

119899

119894= 119904119890

119899minus1

119894⊙ 119888ℎ119886119899119892 119904119890

119899

119894 (16)






119897119890V119890119897 (119904119890119894)119899= 119897119890V119890119897 (119904119890

119894)119899minus1

120573 (17)






















119864is 119878119864119873119878

and 119878119864119873119864





119904119890119894 and 119878119864

119873119878

119904119890119894 and



119904119890119894and 119878119864

119873119864

119904119890119895 then the


(119889119890119897119894V119890119903 119904119890119896 (Ω) = 119878119864

119873119878

119904119890119894⊙ 119878119864119873119864

119904119890119895)

and (V119886119897119906119890 (119889119890119897119894V119890119903 119904119890119896 (Ω)) = V119886119897119906119890 (119878119864

119873119878

119904119890119894)

⊙ V119886119897119906119890 (119878119864119873119864

119904119890119895) 119879 (119873119878 119873119864))

(18)


119904119890119894 and 119878119864119873119864


119889119890119897119894V119890119903 119904119890119896 (Ω) = 119878119864

119873119878

119904119890119894otimes 119878119864119873119864

119904119890119895times Ω (119873

119864 119873119878)

=

119904119890119896119905119910119901119890 = 119891

minus1[119903 (119891 (119904119890

119894) 119891 (119904119890

119895))]

119904119890119896119877119890119897119886119905119890119889 = 119891

minus1(119891 (119904119890

119896119905119910119901119890) 119877119890119897119886119905119894119900119899)

119897119890V119890119897 (119904119890119896) =

119897119890V119890119897 (119904119890119894) + 119897119890V119890119897 (119904119890

119895) minus 1

2

119879 (119873119864 119873119878)

(19)




119889119890119897119894V119890119903 119904119890119896 (Ω) = 119878119864

119873119864

119904119890119895 (20)




119889119890119897119894V119890119903 119904119890119896 (Ω)

=

119889119890119897119894V119890119903 119904119890119896 (Ω) 119905119910119901119890 = 119878119864

119873119878

119904119890119894119905119910119901119890

119889119890119897119894V119890119903 119904119890119896 (Ω) 119877119890119897119886119905119890119889 = 119878119864

119873119878

119904119890119894119877119890119897119886119905119890119889

119889119890119897119894V119890119903 119904119890119896 (Ω) 119897119890V119890119897 = 119878119864119873119878

119904119890119894119897119890V119890119897 times 119879 (119873119878 119873119864)

(21)



119889119890119897119894V119890119903 119878119864 (Ω)

= 119889119890119897119894V119890119903 1199041198901 (Ω) 119889119890119897119894V119890119903 119904119890

2 (Ω)

(22)





119889119890119897119894V119890119903 119878119864119873119864

= 119889119890119897119894V119890119903 119904119890119896(Ω119895) | 119889119890119897119894V119890119903 119904119890

119894(Ω119895)

isin 119889119890119897119894V119890119903 119878119864 (Ω119895) and 119889119890119897119894V119890119903 119878119864 (Ω

119895)

isin⋃119889119890119897119894V119890119903 119878119864 (Ω119895)

(23)


119873119864











119878 119873119864) between two user


119878119864119873119878




generated by 119878119864119873119878


119878 and is

calculated as

119894119898119901119897119894119888119894119905 119904119890119894 (Ω) 119905119910119901119890 = 119878119864

119873119878

119904119890119894119905119910119901119890

119894119898119901119897119894119888119894119905 119904119890119894 (Ω) 119877119890119897119886119905119890119889 = 119878119864

119873119878

119904119890119894119877119890119897119886119905119890119889

119894119898119901119897119894119888119894119905 119904119890119894 (Ω) 119897119890V119890119897 = 119878119864

119873119878

119904119890119894119897119890V119890119897 times 119879 (119873

119878 119873119864)

(24)


119894119898119901119897119894119888119894119905 119878119864 (Ω)

= 119894119898119901119897119894119888119894119905 1199041198901 (Ω) 119894119898119901119897119894119888119894119905 119904119890

2 (Ω)

(25)





119894119898119901119897119894119888119894119905 119878119864119873119864

= 119894119898119901119897119894119888119894119905 119904119890119894(Ω119895) | 119894119898119901119897119894119888119894119905 119904119890

119894(Ω119895)

isin 119894119898119901119897119894119888119894119905 119878119864 (Ω119895) and 119894119898119901119897119894119888119894119905 119878119864 (Ω

119895)

isin⋃ 119894119898119901119897119894119888119894119905 119878119864 (Ω119895)

(26)


119864







119897119890V119890119897(119894119898119901119897119894119888119894119905 119904119890119894)119899

















Accu

racy

of t

rust

chai

n ca

lcul

atio

n (

)

75

80

85

90

95


ATCCTCPTCSTC

(a)

Accu

racy

of t

rust

chai

n ca

lcul

atio

n (

)

70

75

80

85

90

95


ATCCTCPTCSTC

(b)

Accu

racy

of t

rust

chai

n ca

lcul

atio

n (

)

40

50

60

70

80

90

100



(c)






70

80

90

100Av

erag

e acc

urac

y (

)

TCBFAT


6000 8000 10000


40

60

50

70

80

90

100

Aver

age a

ccur

acy

()

TC BFAT


6000 8000 10000


50

60

70

80

90

100

Aver

age a

ccur

acy

()


TCBF

AT



50

60

70

80

90

100Av

erag

e acc

urac

y (

)

TCBF

AT


50

60

70

80

90

100

Aver

age a

ccur

acy

()


TCBF

AT




60

70

80

90

100

Aver

age a

ccur

acy

()

TARTMRSMRCMR


(a)

0

20

40

60

80

100

Aver

age a

ccur

acy

()

20 40 60 80 1000Value of factor 120573

(b)

0

20

40

60

80

100

Accu

racy

()


ESMLDAMESVMKM0

20

40

60

80

100

Accu

racy

()

(d) Finance

ESMLDAMESVMKM0

20

40

60

80

100

Accu

racy

()


0

20

40

60

80

100

Accu

racy

()

(f) Entertainment

Figure 6 Continued


ESMLDAMESVMKM0

20

40

60

80

100

Accu

racy

()

(g) Technology









40

50

60

70

80

90

100Av

erag

e acc

urac

y (

)


ESDBMPEMLE


40

50

60

70

80

90

100

Aver

age a

ccur

acy

()


ESDBMPEMLE


6040 1000 20 800

20

40

60

80

100

Aver

age a

ccur

acy

()


(c)





8 Conclusion



URMTCM

TEENSPFISL0

10

20

30

40

50

60

70Av

erag

e acc

urac

y (

)


URMTCM

TEENSPFISL0

10

20

30

40

50

60

70

Aver

age a

ccur

acy

()


6040 1000 20 800

20

40

60

80

100

Aver

age a

ccur

acy

()


(c)





Acknowledgments


References














































Advances in

FuzzySystems


Volume 2014












Journal of

Journal of





Advances in

Multimedia


Biomedical Imaging



Advances in


RoboticsJournal of










Advances in





119878119874 = ⟨119862119897119886119904119904 119877119900119900119905 ℎ119886119901119901119894119899119890119904119904 119904119886119889119899119890119904119904 119886119899119892119890119903

119877119890119897119886119905119894119900119899 119901119886119903119890119899119905-119888ℎ119894119897119889 ⟨119877119900119900119905 ℎ119886119901119901119894119899119890119904119904⟩

119901119886119903119890119899119905-119888ℎ119894119897119889 ⟨119877119900119900119905 119904119886119889119899119890119904119904⟩

119901119886119903119890119899119905-119888ℎ119894119897119889 ⟨119877119900119900119905 119886119899119892119890119903⟩ 119877119906119897119890 CMR SMR

TMRTARMER⟩

(2)






119894119905119910119901119890 rarr 119878119874119888119897119886119904119904

119895


119894 119888119897119886119904119904

119895) = 119888119897119886119904119904


classes 119888119897119886119904119904119894and 119888119897119886119904119904



119894 119888119897119886119904119904

119895)


SMR 119903 (119905119890119903119903119891119894119890119889 119908119900119903119903119910) = 119889119894119904119905119906119903119887119890119889 (3)




as119897119886119910119890119903 (1198881198971198861199041199042) = 1

119897119886119910119890119903 (1198881198971198861199041199045) = 2

119897119886119910119890119903 (1198881198971198861199041199048) = 3

119891 1199041198901119905119910119901119890 997888rarr 1198781198741198881198971198861199041199042

119891 1199041198902119905119910119901119890 997888rarr 1198781198741198881198971198861199041199045

119891 1199041198904119905119910119901119890 997888rarr 1198781198741198881198971198861199041199048

(4)






1198732and


2isin 1198662)) 119891119894(1198731 1198732)


Level(se4) = 1

Level(se3) = 08

Level(se2) = 08

1Level(se ) = 09

se4

se3

se2

se1


Sentiment dimension




Class 2Class 1

Root




1


1 1198732) can be


119879 (1198731 1198732) =

0 119899 = 0

sum

119899

119894=1(119891119894(1198731 1198732) times 119896119894)

119899

119899 ge 1

(5)





119896119894 =





(6)




119864 and




119879119862(119873119878 119873119864) =

1

119889119890119901119905ℎ (119873119864)

[119879 (119873119878 119873

1

119868)

119889119890119901119905ℎ(119873119864)

+

119889119890119901119905ℎ(119873119864)minus1

sum

119896=2

119879 (119873

119896

119868 119873

119896+1

119868)

119889119890119901119905ℎ(119873119864)minus119896+1

+ 119879 (119873

119889119890119901119905ℎ(119873119864)minus1

119868 119873119864)]

(7)













1



1

119868in 119862119897


119878 1198731



119879 (119873119878 119873119864) =

1

119898

[

[

( sum

119879(119862119897)ge07

119879 (119862119897)11198981

)

+ ( sum

04lt119879(119862119897)lt07

119879 (119862119897)) + ( sum

119879(119862119897)le04

119879 (119862119897)

1198982

)]

]

(8)


1(here 119898



2(here 119898


views)









3) and source node



119879 (1198621) =

1

4

times [1

4+ 09

3+ 07

2+ 08] = 075

119879 (1198622) =

1

3

times [09

3+ 088

2+ 08] = 077

119879 (1198623) =

1

4

times [07

4+ 1

3+ 09

2+ 1] = 076

119879 (1198624) =

1

4

times [08

4+ 09

3+ 1

2+ 07] = 071

119879 (119873119878 119873119864) =

1

4

times [075

13+ 077

13+ 076

13+ 071]

= 086

(9)




B

C

D

E

F

G

H

I

J

M

L

K

A

07

1

07

01

03

08

1

09

09

08

08

109

08

09

07

1

03

03 1

06

09

Source node

Intermediate nodeH

Target node



B

C

D

E

F

G

H

I

J

M

L

K


07

1

07

01

03

08

1

09

09

08

08

109

08

09

07

1

03

03

03 1

06

09

A

Source node

Intermediate nodeH

Target node



K

D

E

F

G

H

J

M

L

KG

Source node

Intermediate nodeH

Target node

A


K

D

E

F

G

H

J

M

L

K


A

Source node

Intermediate nodeH

Target node








1 1199041198902 Then





119894119905119910119901119890 rarr 119878119874119888119897119886119904119904

119894and 119891 119904119890

119895119905119910119901119890 rarr



asV119886119897119906119890 (119904119890

119896) = V119886119897119906119890 (119904119890

119894) ⊙ V119886119897119906119890 (119904119890

119895)

=

(119897119890V119890119897 (119904119890119894) + 119897119886119910 (119888119897119886119904119904119894)) + (119897119890V119890119897 (119904119890119895) + 119897119886119910 (119888119897119886119904119904119895))

2

(10)


as119897119886119910119890119903(119891(119904119890

119896)) = lfloorV119886119897119906119890(119904119890




119896119877119890119897119886119905119890119889 =

119904119890119896119903119890119897119886119905119890119889119896 | 119904119890119896119903119890119897119886119905119890119889119896 = 119891

minus1(119891(119904119890119896119905119910119901119890)119877119890119897119886119905119894119900119899)


V119886119897119906119890 (1199041198905)

=

(119897119890V119890119897 (1199041198901) + 119897119886119910 (119888119897119886119904119904

2)) + (119897119890V119890119897 (119904119890

4) + 119897119886119910 (119888119897119886119904119904

8))

2

=

(1 + 09) + (3 + 1)

2

= 295

(11)


119896119905119910119901119890)) =


6according to


5) minus

119897119886119910119890119903(119891(1199041198905119905119910119901119890)) = 095


119898119904119890 =

119899

⨀

119894=1

119904119890119894 (12)




119894= ⟨119905119910119901119890 119877119890119897119886119905119890119889 119897119890V119890119897⟩ and 119904119890

119895=



SMR is as follows

119904119890119894otimes 119904119890119895= 119904119890119896

=

119904119890119896119905119910119901119890 = 119891

minus1[119903 (119891 (119904119890

119894) 119891 (119904119890

119895))]

119904119890119896119877119890119897119886119905119890119889 = 119891minus1

(119891 (119904119890119896119905119910119901119890) 119877119890119897119886119905119894119900119899)

119897119890V119890119897 (119904119890119896) =

119897119890V119890119897 (119904119890119894) + 119897119890V119890119897 (119904119890119895) minus 1

2

(13)


(exist119891 (1199041198901119905119910119901119890) = 119892119897119886119889 and 119897119890V119890119897 (119904119890

1) ge 07)

and (exist119891 (1199041198902119905119910119901119890) = 119904119886119905119894119904119891119910119894119899119892 and 119897119890V119890119897 (119904119890

2) ge 08)

and (119903 (119892119897119886119889 119904119886119905119894119904119891119910119894119899119892) = ℎ119886119901119901119894119899119890119904119904)

997888rarr (exist (1199041198903119905119910119901119890) = ℎ119886119901119901119894119899119890119904119904 and 119897119890V119890119897 (119904119890

3)

=

119897119890V119890119897 (1199041198901) + 119897119890V119890119897 (119904119890

2) minus 1

2

)

(14)



119894and its next




119904119890

1

119894= 119904119890

0

119894⊙ 119888ℎ119886119899119892 119904119890

1

119894 (15)


119904119890

119899

119894= 119904119890

119899minus1

119894⊙ 119888ℎ119886119899119892 119904119890

119899

119894 (16)






119897119890V119890119897 (119904119890119894)119899= 119897119890V119890119897 (119904119890

119894)119899minus1

120573 (17)






















119864is 119878119864119873119878

and 119878119864119873119864





119904119890119894 and 119878119864

119873119878

119904119890119894 and



119904119890119894and 119878119864

119873119864

119904119890119895 then the


(119889119890119897119894V119890119903 119904119890119896 (Ω) = 119878119864

119873119878

119904119890119894⊙ 119878119864119873119864

119904119890119895)

and (V119886119897119906119890 (119889119890119897119894V119890119903 119904119890119896 (Ω)) = V119886119897119906119890 (119878119864

119873119878

119904119890119894)

⊙ V119886119897119906119890 (119878119864119873119864

119904119890119895) 119879 (119873119878 119873119864))

(18)


119904119890119894 and 119878119864119873119864


119889119890119897119894V119890119903 119904119890119896 (Ω) = 119878119864

119873119878

119904119890119894otimes 119878119864119873119864

119904119890119895times Ω (119873

119864 119873119878)

=

119904119890119896119905119910119901119890 = 119891

minus1[119903 (119891 (119904119890

119894) 119891 (119904119890

119895))]

119904119890119896119877119890119897119886119905119890119889 = 119891

minus1(119891 (119904119890

119896119905119910119901119890) 119877119890119897119886119905119894119900119899)

119897119890V119890119897 (119904119890119896) =

119897119890V119890119897 (119904119890119894) + 119897119890V119890119897 (119904119890

119895) minus 1

2

119879 (119873119864 119873119878)

(19)




119889119890119897119894V119890119903 119904119890119896 (Ω) = 119878119864

119873119864

119904119890119895 (20)




119889119890119897119894V119890119903 119904119890119896 (Ω)

=

119889119890119897119894V119890119903 119904119890119896 (Ω) 119905119910119901119890 = 119878119864

119873119878

119904119890119894119905119910119901119890

119889119890119897119894V119890119903 119904119890119896 (Ω) 119877119890119897119886119905119890119889 = 119878119864

119873119878

119904119890119894119877119890119897119886119905119890119889

119889119890119897119894V119890119903 119904119890119896 (Ω) 119897119890V119890119897 = 119878119864119873119878

119904119890119894119897119890V119890119897 times 119879 (119873119878 119873119864)

(21)



119889119890119897119894V119890119903 119878119864 (Ω)

= 119889119890119897119894V119890119903 1199041198901 (Ω) 119889119890119897119894V119890119903 119904119890

2 (Ω)

(22)





119889119890119897119894V119890119903 119878119864119873119864

= 119889119890119897119894V119890119903 119904119890119896(Ω119895) | 119889119890119897119894V119890119903 119904119890

119894(Ω119895)

isin 119889119890119897119894V119890119903 119878119864 (Ω119895) and 119889119890119897119894V119890119903 119878119864 (Ω

119895)

isin⋃119889119890119897119894V119890119903 119878119864 (Ω119895)

(23)


119873119864











119878 119873119864) between two user


119878119864119873119878




generated by 119878119864119873119878


119878 and is

calculated as

119894119898119901119897119894119888119894119905 119904119890119894 (Ω) 119905119910119901119890 = 119878119864

119873119878

119904119890119894119905119910119901119890

119894119898119901119897119894119888119894119905 119904119890119894 (Ω) 119877119890119897119886119905119890119889 = 119878119864

119873119878

119904119890119894119877119890119897119886119905119890119889

119894119898119901119897119894119888119894119905 119904119890119894 (Ω) 119897119890V119890119897 = 119878119864

119873119878

119904119890119894119897119890V119890119897 times 119879 (119873

119878 119873119864)

(24)


119894119898119901119897119894119888119894119905 119878119864 (Ω)

= 119894119898119901119897119894119888119894119905 1199041198901 (Ω) 119894119898119901119897119894119888119894119905 119904119890

2 (Ω)

(25)





119894119898119901119897119894119888119894119905 119878119864119873119864

= 119894119898119901119897119894119888119894119905 119904119890119894(Ω119895) | 119894119898119901119897119894119888119894119905 119904119890

119894(Ω119895)

isin 119894119898119901119897119894119888119894119905 119878119864 (Ω119895) and 119894119898119901119897119894119888119894119905 119878119864 (Ω

119895)

isin⋃ 119894119898119901119897119894119888119894119905 119878119864 (Ω119895)

(26)


119864







119897119890V119890119897(119894119898119901119897119894119888119894119905 119904119890119894)119899

















Accu

racy

of t

rust

chai

n ca

lcul

atio

n (

)

75

80

85

90

95


ATCCTCPTCSTC

(a)

Accu

racy

of t

rust

chai

n ca

lcul

atio

n (

)

70

75

80

85

90

95


ATCCTCPTCSTC

(b)

Accu

racy

of t

rust

chai

n ca

lcul

atio

n (

)

40

50

60

70

80

90

100



(c)






70

80

90

100Av

erag

e acc

urac

y (

)

TCBFAT


6000 8000 10000


40

60

50

70

80

90

100

Aver

age a

ccur

acy

()

TC BFAT


6000 8000 10000


50

60

70

80

90

100

Aver

age a

ccur

acy

()


TCBF

AT



50

60

70

80

90

100Av

erag

e acc

urac

y (

)

TCBF

AT


50

60

70

80

90

100

Aver

age a

ccur

acy

()


TCBF

AT




60

70

80

90

100

Aver

age a

ccur

acy

()

TARTMRSMRCMR


(a)

0

20

40

60

80

100

Aver

age a

ccur

acy

()

20 40 60 80 1000Value of factor 120573

(b)

0

20

40

60

80

100

Accu

racy

()


ESMLDAMESVMKM0

20

40

60

80

100

Accu

racy

()

(d) Finance

ESMLDAMESVMKM0

20

40

60

80

100

Accu

racy

()


0

20

40

60

80

100

Accu

racy

()

(f) Entertainment

Figure 6 Continued


ESMLDAMESVMKM0

20

40

60

80

100

Accu

racy

()

(g) Technology









40

50

60

70

80

90

100Av

erag

e acc

urac

y (

)


ESDBMPEMLE


40

50

60

70

80

90

100

Aver

age a

ccur

acy

()


ESDBMPEMLE


6040 1000 20 800

20

40

60

80

100

Aver

age a

ccur

acy

()


(c)





8 Conclusion



URMTCM

TEENSPFISL0

10

20

30

40

50

60

70Av

erag

e acc

urac

y (

)


URMTCM

TEENSPFISL0

10

20

30

40

50

60

70

Aver

age a

ccur

acy

()


6040 1000 20 800

20

40

60

80

100

Aver

age a

ccur

acy

()


(c)





Acknowledgments


References














































Advances in

FuzzySystems


Volume 2014












Journal of

Journal of





Advances in

Multimedia


Biomedical Imaging



Advances in


RoboticsJournal of










Advances in




Level(se4) = 1

Level(se3) = 08

Level(se2) = 08

1Level(se ) = 09

se4

se3

se2

se1


Sentiment dimension




Class 2Class 1

Root




1


1 1198732) can be


119879 (1198731 1198732) =

0 119899 = 0

sum

119899

119894=1(119891119894(1198731 1198732) times 119896119894)

119899

119899 ge 1

(5)





119896119894 =





(6)




119864 and




119879119862(119873119878 119873119864) =

1

119889119890119901119905ℎ (119873119864)

[119879 (119873119878 119873

1

119868)

119889119890119901119905ℎ(119873119864)

+

119889119890119901119905ℎ(119873119864)minus1

sum

119896=2

119879 (119873

119896

119868 119873

119896+1

119868)

119889119890119901119905ℎ(119873119864)minus119896+1

+ 119879 (119873

119889119890119901119905ℎ(119873119864)minus1

119868 119873119864)]

(7)













1



1

119868in 119862119897


119878 1198731



119879 (119873119878 119873119864) =

1

119898

[

[

( sum

119879(119862119897)ge07

119879 (119862119897)11198981

)

+ ( sum

04lt119879(119862119897)lt07

119879 (119862119897)) + ( sum

119879(119862119897)le04

119879 (119862119897)

1198982

)]

]

(8)


1(here 119898



2(here 119898


views)









3) and source node



119879 (1198621) =

1

4

times [1

4+ 09

3+ 07

2+ 08] = 075

119879 (1198622) =

1

3

times [09

3+ 088

2+ 08] = 077

119879 (1198623) =

1

4

times [07

4+ 1

3+ 09

2+ 1] = 076

119879 (1198624) =

1

4

times [08

4+ 09

3+ 1

2+ 07] = 071

119879 (119873119878 119873119864) =

1

4

times [075

13+ 077

13+ 076

13+ 071]

= 086

(9)




B

C

D

E

F

G

H

I

J

M

L

K

A

07

1

07

01

03

08

1

09

09

08

08

109

08

09

07

1

03

03 1

06

09

Source node

Intermediate nodeH

Target node



B

C

D

E

F

G

H

I

J

M

L

K


07

1

07

01

03

08

1

09

09

08

08

109

08

09

07

1

03

03

03 1

06

09

A

Source node

Intermediate nodeH

Target node



K

D

E

F

G

H

J

M

L

KG

Source node

Intermediate nodeH

Target node

A


K

D

E

F

G

H

J

M

L

K


A

Source node

Intermediate nodeH

Target node








1 1199041198902 Then





119894119905119910119901119890 rarr 119878119874119888119897119886119904119904

119894and 119891 119904119890

119895119905119910119901119890 rarr



asV119886119897119906119890 (119904119890

119896) = V119886119897119906119890 (119904119890

119894) ⊙ V119886119897119906119890 (119904119890

119895)

=

(119897119890V119890119897 (119904119890119894) + 119897119886119910 (119888119897119886119904119904119894)) + (119897119890V119890119897 (119904119890119895) + 119897119886119910 (119888119897119886119904119904119895))

2

(10)


as119897119886119910119890119903(119891(119904119890

119896)) = lfloorV119886119897119906119890(119904119890




119896119877119890119897119886119905119890119889 =

119904119890119896119903119890119897119886119905119890119889119896 | 119904119890119896119903119890119897119886119905119890119889119896 = 119891

minus1(119891(119904119890119896119905119910119901119890)119877119890119897119886119905119894119900119899)


V119886119897119906119890 (1199041198905)

=

(119897119890V119890119897 (1199041198901) + 119897119886119910 (119888119897119886119904119904

2)) + (119897119890V119890119897 (119904119890

4) + 119897119886119910 (119888119897119886119904119904

8))

2

=

(1 + 09) + (3 + 1)

2

= 295

(11)


119896119905119910119901119890)) =


6according to


5) minus

119897119886119910119890119903(119891(1199041198905119905119910119901119890)) = 095


119898119904119890 =

119899

⨀

119894=1

119904119890119894 (12)




119894= ⟨119905119910119901119890 119877119890119897119886119905119890119889 119897119890V119890119897⟩ and 119904119890

119895=



SMR is as follows

119904119890119894otimes 119904119890119895= 119904119890119896

=

119904119890119896119905119910119901119890 = 119891

minus1[119903 (119891 (119904119890

119894) 119891 (119904119890

119895))]

119904119890119896119877119890119897119886119905119890119889 = 119891minus1

(119891 (119904119890119896119905119910119901119890) 119877119890119897119886119905119894119900119899)

119897119890V119890119897 (119904119890119896) =

119897119890V119890119897 (119904119890119894) + 119897119890V119890119897 (119904119890119895) minus 1

2

(13)


(exist119891 (1199041198901119905119910119901119890) = 119892119897119886119889 and 119897119890V119890119897 (119904119890

1) ge 07)

and (exist119891 (1199041198902119905119910119901119890) = 119904119886119905119894119904119891119910119894119899119892 and 119897119890V119890119897 (119904119890

2) ge 08)

and (119903 (119892119897119886119889 119904119886119905119894119904119891119910119894119899119892) = ℎ119886119901119901119894119899119890119904119904)

997888rarr (exist (1199041198903119905119910119901119890) = ℎ119886119901119901119894119899119890119904119904 and 119897119890V119890119897 (119904119890

3)

=

119897119890V119890119897 (1199041198901) + 119897119890V119890119897 (119904119890

2) minus 1

2

)

(14)



119894and its next




119904119890

1

119894= 119904119890

0

119894⊙ 119888ℎ119886119899119892 119904119890

1

119894 (15)


119904119890

119899

119894= 119904119890

119899minus1

119894⊙ 119888ℎ119886119899119892 119904119890

119899

119894 (16)






119897119890V119890119897 (119904119890119894)119899= 119897119890V119890119897 (119904119890

119894)119899minus1

120573 (17)






















119864is 119878119864119873119878

and 119878119864119873119864





119904119890119894 and 119878119864

119873119878

119904119890119894 and



119904119890119894and 119878119864

119873119864

119904119890119895 then the


(119889119890119897119894V119890119903 119904119890119896 (Ω) = 119878119864

119873119878

119904119890119894⊙ 119878119864119873119864

119904119890119895)

and (V119886119897119906119890 (119889119890119897119894V119890119903 119904119890119896 (Ω)) = V119886119897119906119890 (119878119864

119873119878

119904119890119894)

⊙ V119886119897119906119890 (119878119864119873119864

119904119890119895) 119879 (119873119878 119873119864))

(18)


119904119890119894 and 119878119864119873119864


119889119890119897119894V119890119903 119904119890119896 (Ω) = 119878119864

119873119878

119904119890119894otimes 119878119864119873119864

119904119890119895times Ω (119873

119864 119873119878)

=

119904119890119896119905119910119901119890 = 119891

minus1[119903 (119891 (119904119890

119894) 119891 (119904119890

119895))]

119904119890119896119877119890119897119886119905119890119889 = 119891

minus1(119891 (119904119890

119896119905119910119901119890) 119877119890119897119886119905119894119900119899)

119897119890V119890119897 (119904119890119896) =

119897119890V119890119897 (119904119890119894) + 119897119890V119890119897 (119904119890

119895) minus 1

2

119879 (119873119864 119873119878)

(19)




119889119890119897119894V119890119903 119904119890119896 (Ω) = 119878119864

119873119864

119904119890119895 (20)




119889119890119897119894V119890119903 119904119890119896 (Ω)

=

119889119890119897119894V119890119903 119904119890119896 (Ω) 119905119910119901119890 = 119878119864

119873119878

119904119890119894119905119910119901119890

119889119890119897119894V119890119903 119904119890119896 (Ω) 119877119890119897119886119905119890119889 = 119878119864

119873119878

119904119890119894119877119890119897119886119905119890119889

119889119890119897119894V119890119903 119904119890119896 (Ω) 119897119890V119890119897 = 119878119864119873119878

119904119890119894119897119890V119890119897 times 119879 (119873119878 119873119864)

(21)



119889119890119897119894V119890119903 119878119864 (Ω)

= 119889119890119897119894V119890119903 1199041198901 (Ω) 119889119890119897119894V119890119903 119904119890

2 (Ω)

(22)





119889119890119897119894V119890119903 119878119864119873119864

= 119889119890119897119894V119890119903 119904119890119896(Ω119895) | 119889119890119897119894V119890119903 119904119890

119894(Ω119895)

isin 119889119890119897119894V119890119903 119878119864 (Ω119895) and 119889119890119897119894V119890119903 119878119864 (Ω

119895)

isin⋃119889119890119897119894V119890119903 119878119864 (Ω119895)

(23)


119873119864











119878 119873119864) between two user


119878119864119873119878




generated by 119878119864119873119878


119878 and is

calculated as

119894119898119901119897119894119888119894119905 119904119890119894 (Ω) 119905119910119901119890 = 119878119864

119873119878

119904119890119894119905119910119901119890

119894119898119901119897119894119888119894119905 119904119890119894 (Ω) 119877119890119897119886119905119890119889 = 119878119864

119873119878

119904119890119894119877119890119897119886119905119890119889

119894119898119901119897119894119888119894119905 119904119890119894 (Ω) 119897119890V119890119897 = 119878119864

119873119878

119904119890119894119897119890V119890119897 times 119879 (119873

119878 119873119864)

(24)


119894119898119901119897119894119888119894119905 119878119864 (Ω)

= 119894119898119901119897119894119888119894119905 1199041198901 (Ω) 119894119898119901119897119894119888119894119905 119904119890

2 (Ω)

(25)





119894119898119901119897119894119888119894119905 119878119864119873119864

= 119894119898119901119897119894119888119894119905 119904119890119894(Ω119895) | 119894119898119901119897119894119888119894119905 119904119890

119894(Ω119895)

isin 119894119898119901119897119894119888119894119905 119878119864 (Ω119895) and 119894119898119901119897119894119888119894119905 119878119864 (Ω

119895)

isin⋃ 119894119898119901119897119894119888119894119905 119878119864 (Ω119895)

(26)


119864







119897119890V119890119897(119894119898119901119897119894119888119894119905 119904119890119894)119899

















Accu

racy

of t

rust

chai

n ca

lcul

atio

n (

)

75

80

85

90

95


ATCCTCPTCSTC

(a)

Accu

racy

of t

rust

chai

n ca

lcul

atio

n (

)

70

75

80

85

90

95


ATCCTCPTCSTC

(b)

Accu

racy

of t

rust

chai

n ca

lcul

atio

n (

)

40

50

60

70

80

90

100



(c)






70

80

90

100Av

erag

e acc

urac

y (

)

TCBFAT


6000 8000 10000


40

60

50

70

80

90

100

Aver

age a

ccur

acy

()

TC BFAT


6000 8000 10000


50

60

70

80

90

100

Aver

age a

ccur

acy

()


TCBF

AT



50

60

70

80

90

100Av

erag

e acc

urac

y (

)

TCBF

AT


50

60

70

80

90

100

Aver

age a

ccur

acy

()


TCBF

AT




60

70

80

90

100

Aver

age a

ccur

acy

()

TARTMRSMRCMR


(a)

0

20

40

60

80

100

Aver

age a

ccur

acy

()

20 40 60 80 1000Value of factor 120573

(b)

0

20

40

60

80

100

Accu

racy

()


ESMLDAMESVMKM0

20

40

60

80

100

Accu

racy

()

(d) Finance

ESMLDAMESVMKM0

20

40

60

80

100

Accu

racy

()


0

20

40

60

80

100

Accu

racy

()

(f) Entertainment

Figure 6 Continued


ESMLDAMESVMKM0

20

40

60

80

100

Accu

racy

()

(g) Technology









40

50

60

70

80

90

100Av

erag

e acc

urac

y (

)


ESDBMPEMLE


40

50

60

70

80

90

100

Aver

age a

ccur

acy

()


ESDBMPEMLE


6040 1000 20 800

20

40

60

80

100

Aver

age a

ccur

acy

()


(c)





8 Conclusion



URMTCM

TEENSPFISL0

10

20

30

40

50

60

70Av

erag

e acc

urac

y (

)


URMTCM

TEENSPFISL0

10

20

30

40

50

60

70

Aver

age a

ccur

acy

()


6040 1000 20 800

20

40

60

80

100

Aver

age a

ccur

acy

()


(c)





Acknowledgments


References














































Advances in

FuzzySystems


Volume 2014












Journal of

Journal of





Advances in

Multimedia


Biomedical Imaging



Advances in


RoboticsJournal of










Advances in











1



1

119868in 119862119897


119878 1198731



119879 (119873119878 119873119864) =

1

119898

[

[

( sum

119879(119862119897)ge07

119879 (119862119897)11198981

)

+ ( sum

04lt119879(119862119897)lt07

119879 (119862119897)) + ( sum

119879(119862119897)le04

119879 (119862119897)

1198982

)]

]

(8)


1(here 119898



2(here 119898


views)









3) and source node



119879 (1198621) =

1

4

times [1

4+ 09

3+ 07

2+ 08] = 075

119879 (1198622) =

1

3

times [09

3+ 088

2+ 08] = 077

119879 (1198623) =

1

4

times [07

4+ 1

3+ 09

2+ 1] = 076

119879 (1198624) =

1

4

times [08

4+ 09

3+ 1

2+ 07] = 071

119879 (119873119878 119873119864) =

1

4

times [075

13+ 077

13+ 076

13+ 071]

= 086

(9)




B

C

D

E

F

G

H

I

J

M

L

K

A

07

1

07

01

03

08

1

09

09

08

08

109

08

09

07

1

03

03 1

06

09

Source node

Intermediate nodeH

Target node



B

C

D

E

F

G

H

I

J

M

L

K


07

1

07

01

03

08

1

09

09

08

08

109

08

09

07

1

03

03

03 1

06

09

A

Source node

Intermediate nodeH

Target node



K

D

E

F

G

H

J

M

L

KG

Source node

Intermediate nodeH

Target node

A


K

D

E

F

G

H

J

M

L

K


A

Source node

Intermediate nodeH

Target node








1 1199041198902 Then





119894119905119910119901119890 rarr 119878119874119888119897119886119904119904

119894and 119891 119904119890

119895119905119910119901119890 rarr



asV119886119897119906119890 (119904119890

119896) = V119886119897119906119890 (119904119890

119894) ⊙ V119886119897119906119890 (119904119890

119895)

=

(119897119890V119890119897 (119904119890119894) + 119897119886119910 (119888119897119886119904119904119894)) + (119897119890V119890119897 (119904119890119895) + 119897119886119910 (119888119897119886119904119904119895))

2

(10)


as119897119886119910119890119903(119891(119904119890

119896)) = lfloorV119886119897119906119890(119904119890




119896119877119890119897119886119905119890119889 =

119904119890119896119903119890119897119886119905119890119889119896 | 119904119890119896119903119890119897119886119905119890119889119896 = 119891

minus1(119891(119904119890119896119905119910119901119890)119877119890119897119886119905119894119900119899)


V119886119897119906119890 (1199041198905)

=

(119897119890V119890119897 (1199041198901) + 119897119886119910 (119888119897119886119904119904

2)) + (119897119890V119890119897 (119904119890

4) + 119897119886119910 (119888119897119886119904119904

8))

2

=

(1 + 09) + (3 + 1)

2

= 295

(11)


119896119905119910119901119890)) =


6according to


5) minus

119897119886119910119890119903(119891(1199041198905119905119910119901119890)) = 095


119898119904119890 =

119899

⨀

119894=1

119904119890119894 (12)




119894= ⟨119905119910119901119890 119877119890119897119886119905119890119889 119897119890V119890119897⟩ and 119904119890

119895=



SMR is as follows

119904119890119894otimes 119904119890119895= 119904119890119896

=

119904119890119896119905119910119901119890 = 119891

minus1[119903 (119891 (119904119890

119894) 119891 (119904119890

119895))]

119904119890119896119877119890119897119886119905119890119889 = 119891minus1

(119891 (119904119890119896119905119910119901119890) 119877119890119897119886119905119894119900119899)

119897119890V119890119897 (119904119890119896) =

119897119890V119890119897 (119904119890119894) + 119897119890V119890119897 (119904119890119895) minus 1

2

(13)


(exist119891 (1199041198901119905119910119901119890) = 119892119897119886119889 and 119897119890V119890119897 (119904119890

1) ge 07)

and (exist119891 (1199041198902119905119910119901119890) = 119904119886119905119894119904119891119910119894119899119892 and 119897119890V119890119897 (119904119890

2) ge 08)

and (119903 (119892119897119886119889 119904119886119905119894119904119891119910119894119899119892) = ℎ119886119901119901119894119899119890119904119904)

997888rarr (exist (1199041198903119905119910119901119890) = ℎ119886119901119901119894119899119890119904119904 and 119897119890V119890119897 (119904119890

3)

=

119897119890V119890119897 (1199041198901) + 119897119890V119890119897 (119904119890

2) minus 1

2

)

(14)



119894and its next




119904119890

1

119894= 119904119890

0

119894⊙ 119888ℎ119886119899119892 119904119890

1

119894 (15)


119904119890

119899

119894= 119904119890

119899minus1

119894⊙ 119888ℎ119886119899119892 119904119890

119899

119894 (16)






119897119890V119890119897 (119904119890119894)119899= 119897119890V119890119897 (119904119890

119894)119899minus1

120573 (17)






















119864is 119878119864119873119878

and 119878119864119873119864





119904119890119894 and 119878119864

119873119878

119904119890119894 and



119904119890119894and 119878119864

119873119864

119904119890119895 then the


(119889119890119897119894V119890119903 119904119890119896 (Ω) = 119878119864

119873119878

119904119890119894⊙ 119878119864119873119864

119904119890119895)

and (V119886119897119906119890 (119889119890119897119894V119890119903 119904119890119896 (Ω)) = V119886119897119906119890 (119878119864

119873119878

119904119890119894)

⊙ V119886119897119906119890 (119878119864119873119864

119904119890119895) 119879 (119873119878 119873119864))

(18)


119904119890119894 and 119878119864119873119864


119889119890119897119894V119890119903 119904119890119896 (Ω) = 119878119864

119873119878

119904119890119894otimes 119878119864119873119864

119904119890119895times Ω (119873

119864 119873119878)

=

119904119890119896119905119910119901119890 = 119891

minus1[119903 (119891 (119904119890

119894) 119891 (119904119890

119895))]

119904119890119896119877119890119897119886119905119890119889 = 119891

minus1(119891 (119904119890

119896119905119910119901119890) 119877119890119897119886119905119894119900119899)

119897119890V119890119897 (119904119890119896) =

119897119890V119890119897 (119904119890119894) + 119897119890V119890119897 (119904119890

119895) minus 1

2

119879 (119873119864 119873119878)

(19)




119889119890119897119894V119890119903 119904119890119896 (Ω) = 119878119864

119873119864

119904119890119895 (20)




119889119890119897119894V119890119903 119904119890119896 (Ω)

=

119889119890119897119894V119890119903 119904119890119896 (Ω) 119905119910119901119890 = 119878119864

119873119878

119904119890119894119905119910119901119890

119889119890119897119894V119890119903 119904119890119896 (Ω) 119877119890119897119886119905119890119889 = 119878119864

119873119878

119904119890119894119877119890119897119886119905119890119889

119889119890119897119894V119890119903 119904119890119896 (Ω) 119897119890V119890119897 = 119878119864119873119878

119904119890119894119897119890V119890119897 times 119879 (119873119878 119873119864)

(21)



119889119890119897119894V119890119903 119878119864 (Ω)

= 119889119890119897119894V119890119903 1199041198901 (Ω) 119889119890119897119894V119890119903 119904119890

2 (Ω)

(22)





119889119890119897119894V119890119903 119878119864119873119864

= 119889119890119897119894V119890119903 119904119890119896(Ω119895) | 119889119890119897119894V119890119903 119904119890

119894(Ω119895)

isin 119889119890119897119894V119890119903 119878119864 (Ω119895) and 119889119890119897119894V119890119903 119878119864 (Ω

119895)

isin⋃119889119890119897119894V119890119903 119878119864 (Ω119895)

(23)


119873119864











119878 119873119864) between two user


119878119864119873119878




generated by 119878119864119873119878


119878 and is

calculated as

119894119898119901119897119894119888119894119905 119904119890119894 (Ω) 119905119910119901119890 = 119878119864

119873119878

119904119890119894119905119910119901119890

119894119898119901119897119894119888119894119905 119904119890119894 (Ω) 119877119890119897119886119905119890119889 = 119878119864

119873119878

119904119890119894119877119890119897119886119905119890119889

119894119898119901119897119894119888119894119905 119904119890119894 (Ω) 119897119890V119890119897 = 119878119864

119873119878

119904119890119894119897119890V119890119897 times 119879 (119873

119878 119873119864)

(24)


119894119898119901119897119894119888119894119905 119878119864 (Ω)

= 119894119898119901119897119894119888119894119905 1199041198901 (Ω) 119894119898119901119897119894119888119894119905 119904119890

2 (Ω)

(25)





119894119898119901119897119894119888119894119905 119878119864119873119864

= 119894119898119901119897119894119888119894119905 119904119890119894(Ω119895) | 119894119898119901119897119894119888119894119905 119904119890

119894(Ω119895)

isin 119894119898119901119897119894119888119894119905 119878119864 (Ω119895) and 119894119898119901119897119894119888119894119905 119878119864 (Ω

119895)

isin⋃ 119894119898119901119897119894119888119894119905 119878119864 (Ω119895)

(26)


119864







119897119890V119890119897(119894119898119901119897119894119888119894119905 119904119890119894)119899

















Accu

racy

of t

rust

chai

n ca

lcul

atio

n (

)

75

80

85

90

95


ATCCTCPTCSTC

(a)

Accu

racy

of t

rust

chai

n ca

lcul

atio

n (

)

70

75

80

85

90

95


ATCCTCPTCSTC

(b)

Accu

racy

of t

rust

chai

n ca

lcul

atio

n (

)

40

50

60

70

80

90

100



(c)






70

80

90

100Av

erag

e acc

urac

y (

)

TCBFAT


6000 8000 10000


40

60

50

70

80

90

100

Aver

age a

ccur

acy

()

TC BFAT


6000 8000 10000


50

60

70

80

90

100

Aver

age a

ccur

acy

()


TCBF

AT



50

60

70

80

90

100Av

erag

e acc

urac

y (

)

TCBF

AT


50

60

70

80

90

100

Aver

age a

ccur

acy

()


TCBF

AT




60

70

80

90

100

Aver

age a

ccur

acy

()

TARTMRSMRCMR


(a)

0

20

40

60

80

100

Aver

age a

ccur

acy

()

20 40 60 80 1000Value of factor 120573

(b)

0

20

40

60

80

100

Accu

racy

()


ESMLDAMESVMKM0

20

40

60

80

100

Accu

racy

()

(d) Finance

ESMLDAMESVMKM0

20

40

60

80

100

Accu

racy

()


0

20

40

60

80

100

Accu

racy

()

(f) Entertainment

Figure 6 Continued


ESMLDAMESVMKM0

20

40

60

80

100

Accu

racy

()

(g) Technology









40

50

60

70

80

90

100Av

erag

e acc

urac

y (

)


ESDBMPEMLE


40

50

60

70

80

90

100

Aver

age a

ccur

acy

()


ESDBMPEMLE


6040 1000 20 800

20

40

60

80

100

Aver

age a

ccur

acy

()


(c)





8 Conclusion



URMTCM

TEENSPFISL0

10

20

30

40

50

60

70Av

erag

e acc

urac

y (

)


URMTCM

TEENSPFISL0

10

20

30

40

50

60

70

Aver

age a

ccur

acy

()


6040 1000 20 800

20

40

60

80

100

Aver

age a

ccur

acy

()


(c)





Acknowledgments


References














































Advances in

FuzzySystems


Volume 2014












Journal of

Journal of





Advances in

Multimedia


Biomedical Imaging



Advances in


RoboticsJournal of










Advances in




B

C

D

E

F

G

H

I

J

M

L

K

A

07

1

07

01

03

08

1

09

09

08

08

109

08

09

07

1

03

03 1

06

09

Source node

Intermediate nodeH

Target node



B

C

D

E

F

G

H

I

J

M

L

K


07

1

07

01

03

08

1

09

09

08

08

109

08

09

07

1

03

03

03 1

06

09

A

Source node

Intermediate nodeH

Target node



K

D

E

F

G

H

J

M

L

KG

Source node

Intermediate nodeH

Target node

A


K

D

E

F

G

H

J

M

L

K


A

Source node

Intermediate nodeH

Target node








1 1199041198902 Then





119894119905119910119901119890 rarr 119878119874119888119897119886119904119904

119894and 119891 119904119890

119895119905119910119901119890 rarr



asV119886119897119906119890 (119904119890

119896) = V119886119897119906119890 (119904119890

119894) ⊙ V119886119897119906119890 (119904119890

119895)

=

(119897119890V119890119897 (119904119890119894) + 119897119886119910 (119888119897119886119904119904119894)) + (119897119890V119890119897 (119904119890119895) + 119897119886119910 (119888119897119886119904119904119895))

2

(10)


as119897119886119910119890119903(119891(119904119890

119896)) = lfloorV119886119897119906119890(119904119890




119896119877119890119897119886119905119890119889 =

119904119890119896119903119890119897119886119905119890119889119896 | 119904119890119896119903119890119897119886119905119890119889119896 = 119891

minus1(119891(119904119890119896119905119910119901119890)119877119890119897119886119905119894119900119899)


V119886119897119906119890 (1199041198905)

=

(119897119890V119890119897 (1199041198901) + 119897119886119910 (119888119897119886119904119904

2)) + (119897119890V119890119897 (119904119890

4) + 119897119886119910 (119888119897119886119904119904

8))

2

=

(1 + 09) + (3 + 1)

2

= 295

(11)


119896119905119910119901119890)) =


6according to


5) minus

119897119886119910119890119903(119891(1199041198905119905119910119901119890)) = 095


119898119904119890 =

119899

⨀

119894=1

119904119890119894 (12)




119894= ⟨119905119910119901119890 119877119890119897119886119905119890119889 119897119890V119890119897⟩ and 119904119890

119895=



SMR is as follows

119904119890119894otimes 119904119890119895= 119904119890119896

=

119904119890119896119905119910119901119890 = 119891

minus1[119903 (119891 (119904119890

119894) 119891 (119904119890

119895))]

119904119890119896119877119890119897119886119905119890119889 = 119891minus1

(119891 (119904119890119896119905119910119901119890) 119877119890119897119886119905119894119900119899)

119897119890V119890119897 (119904119890119896) =

119897119890V119890119897 (119904119890119894) + 119897119890V119890119897 (119904119890119895) minus 1

2

(13)


(exist119891 (1199041198901119905119910119901119890) = 119892119897119886119889 and 119897119890V119890119897 (119904119890

1) ge 07)

and (exist119891 (1199041198902119905119910119901119890) = 119904119886119905119894119904119891119910119894119899119892 and 119897119890V119890119897 (119904119890

2) ge 08)

and (119903 (119892119897119886119889 119904119886119905119894119904119891119910119894119899119892) = ℎ119886119901119901119894119899119890119904119904)

997888rarr (exist (1199041198903119905119910119901119890) = ℎ119886119901119901119894119899119890119904119904 and 119897119890V119890119897 (119904119890

3)

=

119897119890V119890119897 (1199041198901) + 119897119890V119890119897 (119904119890

2) minus 1

2

)

(14)



119894and its next




119904119890

1

119894= 119904119890

0

119894⊙ 119888ℎ119886119899119892 119904119890

1

119894 (15)


119904119890

119899

119894= 119904119890

119899minus1

119894⊙ 119888ℎ119886119899119892 119904119890

119899

119894 (16)






119897119890V119890119897 (119904119890119894)119899= 119897119890V119890119897 (119904119890

119894)119899minus1

120573 (17)






















119864is 119878119864119873119878

and 119878119864119873119864





119904119890119894 and 119878119864

119873119878

119904119890119894 and



119904119890119894and 119878119864

119873119864

119904119890119895 then the


(119889119890119897119894V119890119903 119904119890119896 (Ω) = 119878119864

119873119878

119904119890119894⊙ 119878119864119873119864

119904119890119895)

and (V119886119897119906119890 (119889119890119897119894V119890119903 119904119890119896 (Ω)) = V119886119897119906119890 (119878119864

119873119878

119904119890119894)

⊙ V119886119897119906119890 (119878119864119873119864

119904119890119895) 119879 (119873119878 119873119864))

(18)


119904119890119894 and 119878119864119873119864


119889119890119897119894V119890119903 119904119890119896 (Ω) = 119878119864

119873119878

119904119890119894otimes 119878119864119873119864

119904119890119895times Ω (119873

119864 119873119878)

=

119904119890119896119905119910119901119890 = 119891

minus1[119903 (119891 (119904119890

119894) 119891 (119904119890

119895))]

119904119890119896119877119890119897119886119905119890119889 = 119891

minus1(119891 (119904119890

119896119905119910119901119890) 119877119890119897119886119905119894119900119899)

119897119890V119890119897 (119904119890119896) =

119897119890V119890119897 (119904119890119894) + 119897119890V119890119897 (119904119890

119895) minus 1

2

119879 (119873119864 119873119878)

(19)




119889119890119897119894V119890119903 119904119890119896 (Ω) = 119878119864

119873119864

119904119890119895 (20)




119889119890119897119894V119890119903 119904119890119896 (Ω)

=

119889119890119897119894V119890119903 119904119890119896 (Ω) 119905119910119901119890 = 119878119864

119873119878

119904119890119894119905119910119901119890

119889119890119897119894V119890119903 119904119890119896 (Ω) 119877119890119897119886119905119890119889 = 119878119864

119873119878

119904119890119894119877119890119897119886119905119890119889

119889119890119897119894V119890119903 119904119890119896 (Ω) 119897119890V119890119897 = 119878119864119873119878

119904119890119894119897119890V119890119897 times 119879 (119873119878 119873119864)

(21)



119889119890119897119894V119890119903 119878119864 (Ω)

= 119889119890119897119894V119890119903 1199041198901 (Ω) 119889119890119897119894V119890119903 119904119890

2 (Ω)

(22)





119889119890119897119894V119890119903 119878119864119873119864

= 119889119890119897119894V119890119903 119904119890119896(Ω119895) | 119889119890119897119894V119890119903 119904119890

119894(Ω119895)

isin 119889119890119897119894V119890119903 119878119864 (Ω119895) and 119889119890119897119894V119890119903 119878119864 (Ω

119895)

isin⋃119889119890119897119894V119890119903 119878119864 (Ω119895)

(23)


119873119864











119878 119873119864) between two user


119878119864119873119878




generated by 119878119864119873119878


119878 and is

calculated as

119894119898119901119897119894119888119894119905 119904119890119894 (Ω) 119905119910119901119890 = 119878119864

119873119878

119904119890119894119905119910119901119890

119894119898119901119897119894119888119894119905 119904119890119894 (Ω) 119877119890119897119886119905119890119889 = 119878119864

119873119878

119904119890119894119877119890119897119886119905119890119889

119894119898119901119897119894119888119894119905 119904119890119894 (Ω) 119897119890V119890119897 = 119878119864

119873119878

119904119890119894119897119890V119890119897 times 119879 (119873

119878 119873119864)

(24)


119894119898119901119897119894119888119894119905 119878119864 (Ω)

= 119894119898119901119897119894119888119894119905 1199041198901 (Ω) 119894119898119901119897119894119888119894119905 119904119890

2 (Ω)

(25)





119894119898119901119897119894119888119894119905 119878119864119873119864

= 119894119898119901119897119894119888119894119905 119904119890119894(Ω119895) | 119894119898119901119897119894119888119894119905 119904119890

119894(Ω119895)

isin 119894119898119901119897119894119888119894119905 119878119864 (Ω119895) and 119894119898119901119897119894119888119894119905 119878119864 (Ω

119895)

isin⋃ 119894119898119901119897119894119888119894119905 119878119864 (Ω119895)

(26)


119864







119897119890V119890119897(119894119898119901119897119894119888119894119905 119904119890119894)119899

















Accu

racy

of t

rust

chai

n ca

lcul

atio

n (

)

75

80

85

90

95


ATCCTCPTCSTC

(a)

Accu

racy

of t

rust

chai

n ca

lcul

atio

n (

)

70

75

80

85

90

95


ATCCTCPTCSTC

(b)

Accu

racy

of t

rust

chai

n ca

lcul

atio

n (

)

40

50

60

70

80

90

100



(c)






70

80

90

100Av

erag

e acc

urac

y (

)

TCBFAT


6000 8000 10000


40

60

50

70

80

90

100

Aver

age a

ccur

acy

()

TC BFAT


6000 8000 10000


50

60

70

80

90

100

Aver

age a

ccur

acy

()


TCBF

AT



50

60

70

80

90

100Av

erag

e acc

urac

y (

)

TCBF

AT


50

60

70

80

90

100

Aver

age a

ccur

acy

()


TCBF

AT




60

70

80

90

100

Aver

age a

ccur

acy

()

TARTMRSMRCMR


(a)

0

20

40

60

80

100

Aver

age a

ccur

acy

()

20 40 60 80 1000Value of factor 120573

(b)

0

20

40

60

80

100

Accu

racy

()


ESMLDAMESVMKM0

20

40

60

80

100

Accu

racy

()

(d) Finance

ESMLDAMESVMKM0

20

40

60

80

100

Accu

racy

()


0

20

40

60

80

100

Accu

racy

()

(f) Entertainment

Figure 6 Continued


ESMLDAMESVMKM0

20

40

60

80

100

Accu

racy

()

(g) Technology









40

50

60

70

80

90

100Av

erag

e acc

urac

y (

)


ESDBMPEMLE


40

50

60

70

80

90

100

Aver

age a

ccur

acy

()


ESDBMPEMLE


6040 1000 20 800

20

40

60

80

100

Aver

age a

ccur

acy

()


(c)





8 Conclusion



URMTCM

TEENSPFISL0

10

20

30

40

50

60

70Av

erag

e acc

urac

y (

)


URMTCM

TEENSPFISL0

10

20

30

40

50

60

70

Aver

age a

ccur

acy

()


6040 1000 20 800

20

40

60

80

100

Aver

age a

ccur

acy

()


(c)





Acknowledgments


References














































Advances in

FuzzySystems


Volume 2014












Journal of

Journal of





Advances in

Multimedia


Biomedical Imaging



Advances in


RoboticsJournal of










Advances in





1 1199041198902 Then





119894119905119910119901119890 rarr 119878119874119888119897119886119904119904

119894and 119891 119904119890

119895119905119910119901119890 rarr



asV119886119897119906119890 (119904119890

119896) = V119886119897119906119890 (119904119890

119894) ⊙ V119886119897119906119890 (119904119890

119895)

=

(119897119890V119890119897 (119904119890119894) + 119897119886119910 (119888119897119886119904119904119894)) + (119897119890V119890119897 (119904119890119895) + 119897119886119910 (119888119897119886119904119904119895))

2

(10)


as119897119886119910119890119903(119891(119904119890

119896)) = lfloorV119886119897119906119890(119904119890




119896119877119890119897119886119905119890119889 =

119904119890119896119903119890119897119886119905119890119889119896 | 119904119890119896119903119890119897119886119905119890119889119896 = 119891

minus1(119891(119904119890119896119905119910119901119890)119877119890119897119886119905119894119900119899)


V119886119897119906119890 (1199041198905)

=

(119897119890V119890119897 (1199041198901) + 119897119886119910 (119888119897119886119904119904

2)) + (119897119890V119890119897 (119904119890

4) + 119897119886119910 (119888119897119886119904119904

8))

2

=

(1 + 09) + (3 + 1)

2

= 295

(11)


119896119905119910119901119890)) =


6according to


5) minus

119897119886119910119890119903(119891(1199041198905119905119910119901119890)) = 095


119898119904119890 =

119899

⨀

119894=1

119904119890119894 (12)




119894= ⟨119905119910119901119890 119877119890119897119886119905119890119889 119897119890V119890119897⟩ and 119904119890

119895=



SMR is as follows

119904119890119894otimes 119904119890119895= 119904119890119896

=

119904119890119896119905119910119901119890 = 119891

minus1[119903 (119891 (119904119890

119894) 119891 (119904119890

119895))]

119904119890119896119877119890119897119886119905119890119889 = 119891minus1

(119891 (119904119890119896119905119910119901119890) 119877119890119897119886119905119894119900119899)

119897119890V119890119897 (119904119890119896) =

119897119890V119890119897 (119904119890119894) + 119897119890V119890119897 (119904119890119895) minus 1

2

(13)


(exist119891 (1199041198901119905119910119901119890) = 119892119897119886119889 and 119897119890V119890119897 (119904119890

1) ge 07)

and (exist119891 (1199041198902119905119910119901119890) = 119904119886119905119894119904119891119910119894119899119892 and 119897119890V119890119897 (119904119890

2) ge 08)

and (119903 (119892119897119886119889 119904119886119905119894119904119891119910119894119899119892) = ℎ119886119901119901119894119899119890119904119904)

997888rarr (exist (1199041198903119905119910119901119890) = ℎ119886119901119901119894119899119890119904119904 and 119897119890V119890119897 (119904119890

3)

=

119897119890V119890119897 (1199041198901) + 119897119890V119890119897 (119904119890

2) minus 1

2

)

(14)



119894and its next




119904119890

1

119894= 119904119890

0

119894⊙ 119888ℎ119886119899119892 119904119890

1

119894 (15)


119904119890

119899

119894= 119904119890

119899minus1

119894⊙ 119888ℎ119886119899119892 119904119890

119899

119894 (16)






119897119890V119890119897 (119904119890119894)119899= 119897119890V119890119897 (119904119890

119894)119899minus1

120573 (17)






















119864is 119878119864119873119878

and 119878119864119873119864





119904119890119894 and 119878119864

119873119878

119904119890119894 and



119904119890119894and 119878119864

119873119864

119904119890119895 then the


(119889119890119897119894V119890119903 119904119890119896 (Ω) = 119878119864

119873119878

119904119890119894⊙ 119878119864119873119864

119904119890119895)

and (V119886119897119906119890 (119889119890119897119894V119890119903 119904119890119896 (Ω)) = V119886119897119906119890 (119878119864

119873119878

119904119890119894)

⊙ V119886119897119906119890 (119878119864119873119864

119904119890119895) 119879 (119873119878 119873119864))

(18)


119904119890119894 and 119878119864119873119864


119889119890119897119894V119890119903 119904119890119896 (Ω) = 119878119864

119873119878

119904119890119894otimes 119878119864119873119864

119904119890119895times Ω (119873

119864 119873119878)

=

119904119890119896119905119910119901119890 = 119891

minus1[119903 (119891 (119904119890

119894) 119891 (119904119890

119895))]

119904119890119896119877119890119897119886119905119890119889 = 119891

minus1(119891 (119904119890

119896119905119910119901119890) 119877119890119897119886119905119894119900119899)

119897119890V119890119897 (119904119890119896) =

119897119890V119890119897 (119904119890119894) + 119897119890V119890119897 (119904119890

119895) minus 1

2

119879 (119873119864 119873119878)

(19)




119889119890119897119894V119890119903 119904119890119896 (Ω) = 119878119864

119873119864

119904119890119895 (20)




119889119890119897119894V119890119903 119904119890119896 (Ω)

=

119889119890119897119894V119890119903 119904119890119896 (Ω) 119905119910119901119890 = 119878119864

119873119878

119904119890119894119905119910119901119890

119889119890119897119894V119890119903 119904119890119896 (Ω) 119877119890119897119886119905119890119889 = 119878119864

119873119878

119904119890119894119877119890119897119886119905119890119889

119889119890119897119894V119890119903 119904119890119896 (Ω) 119897119890V119890119897 = 119878119864119873119878

119904119890119894119897119890V119890119897 times 119879 (119873119878 119873119864)

(21)



119889119890119897119894V119890119903 119878119864 (Ω)

= 119889119890119897119894V119890119903 1199041198901 (Ω) 119889119890119897119894V119890119903 119904119890

2 (Ω)

(22)





119889119890119897119894V119890119903 119878119864119873119864

= 119889119890119897119894V119890119903 119904119890119896(Ω119895) | 119889119890119897119894V119890119903 119904119890

119894(Ω119895)

isin 119889119890119897119894V119890119903 119878119864 (Ω119895) and 119889119890119897119894V119890119903 119878119864 (Ω

119895)

isin⋃119889119890119897119894V119890119903 119878119864 (Ω119895)

(23)


119873119864











119878 119873119864) between two user


119878119864119873119878




generated by 119878119864119873119878


119878 and is

calculated as

119894119898119901119897119894119888119894119905 119904119890119894 (Ω) 119905119910119901119890 = 119878119864

119873119878

119904119890119894119905119910119901119890

119894119898119901119897119894119888119894119905 119904119890119894 (Ω) 119877119890119897119886119905119890119889 = 119878119864

119873119878

119904119890119894119877119890119897119886119905119890119889

119894119898119901119897119894119888119894119905 119904119890119894 (Ω) 119897119890V119890119897 = 119878119864

119873119878

119904119890119894119897119890V119890119897 times 119879 (119873

119878 119873119864)

(24)


119894119898119901119897119894119888119894119905 119878119864 (Ω)

= 119894119898119901119897119894119888119894119905 1199041198901 (Ω) 119894119898119901119897119894119888119894119905 119904119890

2 (Ω)

(25)





119894119898119901119897119894119888119894119905 119878119864119873119864

= 119894119898119901119897119894119888119894119905 119904119890119894(Ω119895) | 119894119898119901119897119894119888119894119905 119904119890

119894(Ω119895)

isin 119894119898119901119897119894119888119894119905 119878119864 (Ω119895) and 119894119898119901119897119894119888119894119905 119878119864 (Ω

119895)

isin⋃ 119894119898119901119897119894119888119894119905 119878119864 (Ω119895)

(26)


119864







119897119890V119890119897(119894119898119901119897119894119888119894119905 119904119890119894)119899

















Accu

racy

of t

rust

chai

n ca

lcul

atio

n (

)

75

80

85

90

95


ATCCTCPTCSTC

(a)

Accu

racy

of t

rust

chai

n ca

lcul

atio

n (

)

70

75

80

85

90

95


ATCCTCPTCSTC

(b)

Accu

racy

of t

rust

chai

n ca

lcul

atio

n (

)

40

50

60

70

80

90

100



(c)






70

80

90

100Av

erag

e acc

urac

y (

)

TCBFAT


6000 8000 10000


40

60

50

70

80

90

100

Aver

age a

ccur

acy

()

TC BFAT


6000 8000 10000


50

60

70

80

90

100

Aver

age a

ccur

acy

()


TCBF

AT



50

60

70

80

90

100Av

erag

e acc

urac

y (

)

TCBF

AT


50

60

70

80

90

100

Aver

age a

ccur

acy

()


TCBF

AT




60

70

80

90

100

Aver

age a

ccur

acy

()

TARTMRSMRCMR


(a)

0

20

40

60

80

100

Aver

age a

ccur

acy

()

20 40 60 80 1000Value of factor 120573

(b)

0

20

40

60

80

100

Accu

racy

()


ESMLDAMESVMKM0

20

40

60

80

100

Accu

racy

()

(d) Finance

ESMLDAMESVMKM0

20

40

60

80

100

Accu

racy

()


0

20

40

60

80

100

Accu

racy

()

(f) Entertainment

Figure 6 Continued


ESMLDAMESVMKM0

20

40

60

80

100

Accu

racy

()

(g) Technology









40

50

60

70

80

90

100Av

erag

e acc

urac

y (

)


ESDBMPEMLE


40

50

60

70

80

90

100

Aver

age a

ccur

acy

()


ESDBMPEMLE


6040 1000 20 800

20

40

60

80

100

Aver

age a

ccur

acy

()


(c)





8 Conclusion



URMTCM

TEENSPFISL0

10

20

30

40

50

60

70Av

erag

e acc

urac

y (

)


URMTCM

TEENSPFISL0

10

20

30

40

50

60

70

Aver

age a

ccur

acy

()


6040 1000 20 800

20

40

60

80

100

Aver

age a

ccur

acy

()


(c)





Acknowledgments


References














































Advances in

FuzzySystems


Volume 2014












Journal of

Journal of





Advances in

Multimedia


Biomedical Imaging



Advances in


RoboticsJournal of










Advances in





119897119890V119890119897 (119904119890119894)119899= 119897119890V119890119897 (119904119890

119894)119899minus1

120573 (17)






















119864is 119878119864119873119878

and 119878119864119873119864





119904119890119894 and 119878119864

119873119878

119904119890119894 and



119904119890119894and 119878119864

119873119864

119904119890119895 then the


(119889119890119897119894V119890119903 119904119890119896 (Ω) = 119878119864

119873119878

119904119890119894⊙ 119878119864119873119864

119904119890119895)

and (V119886119897119906119890 (119889119890119897119894V119890119903 119904119890119896 (Ω)) = V119886119897119906119890 (119878119864

119873119878

119904119890119894)

⊙ V119886119897119906119890 (119878119864119873119864

119904119890119895) 119879 (119873119878 119873119864))

(18)


119904119890119894 and 119878119864119873119864


119889119890119897119894V119890119903 119904119890119896 (Ω) = 119878119864

119873119878

119904119890119894otimes 119878119864119873119864

119904119890119895times Ω (119873

119864 119873119878)

=

119904119890119896119905119910119901119890 = 119891

minus1[119903 (119891 (119904119890

119894) 119891 (119904119890

119895))]

119904119890119896119877119890119897119886119905119890119889 = 119891

minus1(119891 (119904119890

119896119905119910119901119890) 119877119890119897119886119905119894119900119899)

119897119890V119890119897 (119904119890119896) =

119897119890V119890119897 (119904119890119894) + 119897119890V119890119897 (119904119890

119895) minus 1

2

119879 (119873119864 119873119878)

(19)




119889119890119897119894V119890119903 119904119890119896 (Ω) = 119878119864

119873119864

119904119890119895 (20)




119889119890119897119894V119890119903 119904119890119896 (Ω)

=

119889119890119897119894V119890119903 119904119890119896 (Ω) 119905119910119901119890 = 119878119864

119873119878

119904119890119894119905119910119901119890

119889119890119897119894V119890119903 119904119890119896 (Ω) 119877119890119897119886119905119890119889 = 119878119864

119873119878

119904119890119894119877119890119897119886119905119890119889

119889119890119897119894V119890119903 119904119890119896 (Ω) 119897119890V119890119897 = 119878119864119873119878

119904119890119894119897119890V119890119897 times 119879 (119873119878 119873119864)

(21)



119889119890119897119894V119890119903 119878119864 (Ω)

= 119889119890119897119894V119890119903 1199041198901 (Ω) 119889119890119897119894V119890119903 119904119890

2 (Ω)

(22)





119889119890119897119894V119890119903 119878119864119873119864

= 119889119890119897119894V119890119903 119904119890119896(Ω119895) | 119889119890119897119894V119890119903 119904119890

119894(Ω119895)

isin 119889119890119897119894V119890119903 119878119864 (Ω119895) and 119889119890119897119894V119890119903 119878119864 (Ω

119895)

isin⋃119889119890119897119894V119890119903 119878119864 (Ω119895)

(23)


119873119864











119878 119873119864) between two user


119878119864119873119878




generated by 119878119864119873119878


119878 and is

calculated as

119894119898119901119897119894119888119894119905 119904119890119894 (Ω) 119905119910119901119890 = 119878119864

119873119878

119904119890119894119905119910119901119890

119894119898119901119897119894119888119894119905 119904119890119894 (Ω) 119877119890119897119886119905119890119889 = 119878119864

119873119878

119904119890119894119877119890119897119886119905119890119889

119894119898119901119897119894119888119894119905 119904119890119894 (Ω) 119897119890V119890119897 = 119878119864

119873119878

119904119890119894119897119890V119890119897 times 119879 (119873

119878 119873119864)

(24)


119894119898119901119897119894119888119894119905 119878119864 (Ω)

= 119894119898119901119897119894119888119894119905 1199041198901 (Ω) 119894119898119901119897119894119888119894119905 119904119890

2 (Ω)

(25)





119894119898119901119897119894119888119894119905 119878119864119873119864

= 119894119898119901119897119894119888119894119905 119904119890119894(Ω119895) | 119894119898119901119897119894119888119894119905 119904119890

119894(Ω119895)

isin 119894119898119901119897119894119888119894119905 119878119864 (Ω119895) and 119894119898119901119897119894119888119894119905 119878119864 (Ω

119895)

isin⋃ 119894119898119901119897119894119888119894119905 119878119864 (Ω119895)

(26)


119864







119897119890V119890119897(119894119898119901119897119894119888119894119905 119904119890119894)119899

















Accu

racy

of t

rust

chai

n ca

lcul

atio

n (

)

75

80

85

90

95


ATCCTCPTCSTC

(a)

Accu

racy

of t

rust

chai

n ca

lcul

atio

n (

)

70

75

80

85

90

95


ATCCTCPTCSTC

(b)

Accu

racy

of t

rust

chai

n ca

lcul

atio

n (

)

40

50

60

70

80

90

100



(c)






70

80

90

100Av

erag

e acc

urac

y (

)

TCBFAT


6000 8000 10000


40

60

50

70

80

90

100

Aver

age a

ccur

acy

()

TC BFAT


6000 8000 10000


50

60

70

80

90

100

Aver

age a

ccur

acy

()


TCBF

AT



50

60

70

80

90

100Av

erag

e acc

urac

y (

)

TCBF

AT


50

60

70

80

90

100

Aver

age a

ccur

acy

()


TCBF

AT




60

70

80

90

100

Aver

age a

ccur

acy

()

TARTMRSMRCMR


(a)

0

20

40

60

80

100

Aver

age a

ccur

acy

()

20 40 60 80 1000Value of factor 120573

(b)

0

20

40

60

80

100

Accu

racy

()


ESMLDAMESVMKM0

20

40

60

80

100

Accu

racy

()

(d) Finance

ESMLDAMESVMKM0

20

40

60

80

100

Accu

racy

()


0

20

40

60

80

100

Accu

racy

()

(f) Entertainment

Figure 6 Continued


ESMLDAMESVMKM0

20

40

60

80

100

Accu

racy

()

(g) Technology









40

50

60

70

80

90

100Av

erag

e acc

urac

y (

)


ESDBMPEMLE


40

50

60

70

80

90

100

Aver

age a

ccur

acy

()


ESDBMPEMLE


6040 1000 20 800

20

40

60

80

100

Aver

age a

ccur

acy

()


(c)





8 Conclusion



URMTCM

TEENSPFISL0

10

20

30

40

50

60

70Av

erag

e acc

urac

y (

)


URMTCM

TEENSPFISL0

10

20

30

40

50

60

70

Aver

age a

ccur

acy

()


6040 1000 20 800

20

40

60

80

100

Aver

age a

ccur

acy

()


(c)





Acknowledgments


References














































Advances in

FuzzySystems


Volume 2014












Journal of

Journal of





Advances in

Multimedia


Biomedical Imaging



Advances in


RoboticsJournal of










Advances in





119889119890119897119894V119890119903 119878119864 (Ω)

= 119889119890119897119894V119890119903 1199041198901 (Ω) 119889119890119897119894V119890119903 119904119890

2 (Ω)

(22)





119889119890119897119894V119890119903 119878119864119873119864

= 119889119890119897119894V119890119903 119904119890119896(Ω119895) | 119889119890119897119894V119890119903 119904119890

119894(Ω119895)

isin 119889119890119897119894V119890119903 119878119864 (Ω119895) and 119889119890119897119894V119890119903 119878119864 (Ω

119895)

isin⋃119889119890119897119894V119890119903 119878119864 (Ω119895)

(23)


119873119864











119878 119873119864) between two user


119878119864119873119878




generated by 119878119864119873119878


119878 and is

calculated as

119894119898119901119897119894119888119894119905 119904119890119894 (Ω) 119905119910119901119890 = 119878119864

119873119878

119904119890119894119905119910119901119890

119894119898119901119897119894119888119894119905 119904119890119894 (Ω) 119877119890119897119886119905119890119889 = 119878119864

119873119878

119904119890119894119877119890119897119886119905119890119889

119894119898119901119897119894119888119894119905 119904119890119894 (Ω) 119897119890V119890119897 = 119878119864

119873119878

119904119890119894119897119890V119890119897 times 119879 (119873

119878 119873119864)

(24)


119894119898119901119897119894119888119894119905 119878119864 (Ω)

= 119894119898119901119897119894119888119894119905 1199041198901 (Ω) 119894119898119901119897119894119888119894119905 119904119890

2 (Ω)

(25)





119894119898119901119897119894119888119894119905 119878119864119873119864

= 119894119898119901119897119894119888119894119905 119904119890119894(Ω119895) | 119894119898119901119897119894119888119894119905 119904119890

119894(Ω119895)

isin 119894119898119901119897119894119888119894119905 119878119864 (Ω119895) and 119894119898119901119897119894119888119894119905 119878119864 (Ω

119895)

isin⋃ 119894119898119901119897119894119888119894119905 119878119864 (Ω119895)

(26)


119864







119897119890V119890119897(119894119898119901119897119894119888119894119905 119904119890119894)119899

















Accu

racy

of t

rust

chai

n ca

lcul

atio

n (

)

75

80

85

90

95


ATCCTCPTCSTC

(a)

Accu

racy

of t

rust

chai

n ca

lcul

atio

n (

)

70

75

80

85

90

95


ATCCTCPTCSTC

(b)

Accu

racy

of t

rust

chai

n ca

lcul

atio

n (

)

40

50

60

70

80

90

100



(c)






70

80

90

100Av

erag

e acc

urac

y (

)

TCBFAT


6000 8000 10000


40

60

50

70

80

90

100

Aver

age a

ccur

acy

()

TC BFAT


6000 8000 10000


50

60

70

80

90

100

Aver

age a

ccur

acy

()


TCBF

AT



50

60

70

80

90

100Av

erag

e acc

urac

y (

)

TCBF

AT


50

60

70

80

90

100

Aver

age a

ccur

acy

()


TCBF

AT




60

70

80

90

100

Aver

age a

ccur

acy

()

TARTMRSMRCMR


(a)

0

20

40

60

80

100

Aver

age a

ccur

acy

()

20 40 60 80 1000Value of factor 120573

(b)

0

20

40

60

80

100

Accu

racy

()


ESMLDAMESVMKM0

20

40

60

80

100

Accu

racy

()

(d) Finance

ESMLDAMESVMKM0

20

40

60

80

100

Accu

racy

()


0

20

40

60

80

100

Accu

racy

()

(f) Entertainment

Figure 6 Continued


ESMLDAMESVMKM0

20

40

60

80

100

Accu

racy

()

(g) Technology









40

50

60

70

80

90

100Av

erag

e acc

urac

y (

)


ESDBMPEMLE


40

50

60

70

80

90

100

Aver

age a

ccur

acy

()


ESDBMPEMLE


6040 1000 20 800

20

40

60

80

100

Aver

age a

ccur

acy

()


(c)





8 Conclusion



URMTCM

TEENSPFISL0

10

20

30

40

50

60

70Av

erag

e acc

urac

y (

)


URMTCM

TEENSPFISL0

10

20

30

40

50

60

70

Aver

age a

ccur

acy

()


6040 1000 20 800

20

40

60

80

100

Aver

age a

ccur

acy

()


(c)





Acknowledgments


References














































Advances in

FuzzySystems


Volume 2014












Journal of

Journal of





Advances in

Multimedia


Biomedical Imaging



Advances in


RoboticsJournal of










Advances in
















Accu

racy

of t

rust

chai

n ca

lcul

atio

n (

)

75

80

85

90

95


ATCCTCPTCSTC

(a)

Accu

racy

of t

rust

chai

n ca

lcul

atio

n (

)

70

75

80

85

90

95


ATCCTCPTCSTC

(b)

Accu

racy

of t

rust

chai

n ca

lcul

atio

n (

)

40

50

60

70

80

90

100



(c)






70

80

90

100Av

erag

e acc

urac

y (

)

TCBFAT


6000 8000 10000


40

60

50

70

80

90

100

Aver

age a

ccur

acy

()

TC BFAT


6000 8000 10000


50

60

70

80

90

100

Aver

age a

ccur

acy

()


TCBF

AT



50

60

70

80

90

100Av

erag

e acc

urac

y (

)

TCBF

AT


50

60

70

80

90

100

Aver

age a

ccur

acy

()


TCBF

AT




60

70

80

90

100

Aver

age a

ccur

acy

()

TARTMRSMRCMR


(a)

0

20

40

60

80

100

Aver

age a

ccur

acy

()

20 40 60 80 1000Value of factor 120573

(b)

0

20

40

60

80

100

Accu

racy

()


ESMLDAMESVMKM0

20

40

60

80

100

Accu

racy

()

(d) Finance

ESMLDAMESVMKM0

20

40

60

80

100

Accu

racy

()


0

20

40

60

80

100

Accu

racy

()

(f) Entertainment

Figure 6 Continued


ESMLDAMESVMKM0

20

40

60

80

100

Accu

racy

()

(g) Technology









40

50

60

70

80

90

100Av

erag

e acc

urac

y (

)


ESDBMPEMLE


40

50

60

70

80

90

100

Aver

age a

ccur

acy

()


ESDBMPEMLE


6040 1000 20 800

20

40

60

80

100

Aver

age a

ccur

acy

()


(c)





8 Conclusion



URMTCM

TEENSPFISL0

10

20

30

40

50

60

70Av

erag

e acc

urac

y (

)


URMTCM

TEENSPFISL0

10

20

30

40

50

60

70

Aver

age a

ccur

acy

()


6040 1000 20 800

20

40

60

80

100

Aver

age a

ccur

acy

()


(c)





Acknowledgments


References














































Advances in

FuzzySystems


Volume 2014












Journal of

Journal of





Advances in

Multimedia


Biomedical Imaging



Advances in


RoboticsJournal of










Advances in




Accu

racy

of t

rust

chai

n ca

lcul

atio

n (

)

75

80

85

90

95


ATCCTCPTCSTC

(a)

Accu

racy

of t

rust

chai

n ca

lcul

atio

n (

)

70

75

80

85

90

95


ATCCTCPTCSTC

(b)

Accu

racy

of t

rust

chai

n ca

lcul

atio

n (

)

40

50

60

70

80

90

100



(c)






70

80

90

100Av

erag

e acc

urac

y (

)

TCBFAT


6000 8000 10000


40

60

50

70

80

90

100

Aver

age a

ccur

acy

()

TC BFAT


6000 8000 10000


50

60

70

80

90

100

Aver

age a

ccur

acy

()


TCBF

AT



50

60

70

80

90

100Av

erag

e acc

urac

y (

)

TCBF

AT


50

60

70

80

90

100

Aver

age a

ccur

acy

()


TCBF

AT




60

70

80

90

100

Aver

age a

ccur

acy

()

TARTMRSMRCMR


(a)

0

20

40

60

80

100

Aver

age a

ccur

acy

()

20 40 60 80 1000Value of factor 120573

(b)

0

20

40

60

80

100

Accu

racy

()


ESMLDAMESVMKM0

20

40

60

80

100

Accu

racy

()

(d) Finance

ESMLDAMESVMKM0

20

40

60

80

100

Accu

racy

()


0

20

40

60

80

100

Accu

racy

()

(f) Entertainment

Figure 6 Continued


ESMLDAMESVMKM0

20

40

60

80

100

Accu

racy

()

(g) Technology









40

50

60

70

80

90

100Av

erag

e acc

urac

y (

)


ESDBMPEMLE


40

50

60

70

80

90

100

Aver

age a

ccur

acy

()


ESDBMPEMLE


6040 1000 20 800

20

40

60

80

100

Aver

age a

ccur

acy

()


(c)





8 Conclusion



URMTCM

TEENSPFISL0

10

20

30

40

50

60

70Av

erag

e acc

urac

y (

)


URMTCM

TEENSPFISL0

10

20

30

40

50

60

70

Aver

age a

ccur

acy

()


6040 1000 20 800

20

40

60

80

100

Aver

age a

ccur

acy

()


(c)





Acknowledgments


References














































Advances in

FuzzySystems


Volume 2014












Journal of

Journal of





Advances in

Multimedia


Biomedical Imaging



Advances in


RoboticsJournal of










Advances in




70

80

90

100Av

erag

e acc

urac

y (

)

TCBFAT


6000 8000 10000


40

60

50

70

80

90

100

Aver

age a

ccur

acy

()

TC BFAT


6000 8000 10000


50

60

70

80

90

100

Aver

age a

ccur

acy

()


TCBF

AT



50

60

70

80

90

100Av

erag

e acc

urac

y (

)

TCBF

AT


50

60

70

80

90

100

Aver

age a

ccur

acy

()


TCBF

AT




60

70

80

90

100

Aver

age a

ccur

acy

()

TARTMRSMRCMR


(a)

0

20

40

60

80

100

Aver

age a

ccur

acy

()

20 40 60 80 1000Value of factor 120573

(b)

0

20

40

60

80

100

Accu

racy

()


ESMLDAMESVMKM0

20

40

60

80

100

Accu

racy

()

(d) Finance

ESMLDAMESVMKM0

20

40

60

80

100

Accu

racy

()


0

20

40

60

80

100

Accu

racy

()

(f) Entertainment

Figure 6 Continued


ESMLDAMESVMKM0

20

40

60

80

100

Accu

racy

()

(g) Technology









40

50

60

70

80

90

100Av

erag

e acc

urac

y (

)


ESDBMPEMLE


40

50

60

70

80

90

100

Aver

age a

ccur

acy

()


ESDBMPEMLE


6040 1000 20 800

20

40

60

80

100

Aver

age a

ccur

acy

()


(c)





8 Conclusion



URMTCM

TEENSPFISL0

10

20

30

40

50

60

70Av

erag

e acc

urac

y (

)


URMTCM

TEENSPFISL0

10

20

30

40

50

60

70

Aver

age a

ccur

acy

()


6040 1000 20 800

20

40

60

80

100

Aver

age a

ccur

acy

()


(c)





Acknowledgments


References














































Advances in

FuzzySystems


Volume 2014












Journal of

Journal of





Advances in

Multimedia


Biomedical Imaging



Advances in


RoboticsJournal of










Advances in




60

70

80

90

100

Aver

age a

ccur

acy

()

TARTMRSMRCMR


(a)

0

20

40

60

80

100

Aver

age a

ccur

acy

()

20 40 60 80 1000Value of factor 120573

(b)

0

20

40

60

80

100

Accu

racy

()


ESMLDAMESVMKM0

20

40

60

80

100

Accu

racy

()

(d) Finance

ESMLDAMESVMKM0

20

40

60

80

100

Accu

racy

()


0

20

40

60

80

100

Accu

racy

()

(f) Entertainment

Figure 6 Continued


ESMLDAMESVMKM0

20

40

60

80

100

Accu

racy

()

(g) Technology









40

50

60

70

80

90

100Av

erag

e acc

urac

y (

)


ESDBMPEMLE


40

50

60

70

80

90

100

Aver

age a

ccur

acy

()


ESDBMPEMLE


6040 1000 20 800

20

40

60

80

100

Aver

age a

ccur

acy

()


(c)





8 Conclusion



URMTCM

TEENSPFISL0

10

20

30

40

50

60

70Av

erag

e acc

urac

y (

)


URMTCM

TEENSPFISL0

10

20

30

40

50

60

70

Aver

age a

ccur

acy

()


6040 1000 20 800

20

40

60

80

100

Aver

age a

ccur

acy

()


(c)





Acknowledgments


References














































Advances in

FuzzySystems


Volume 2014












Journal of

Journal of





Advances in

Multimedia


Biomedical Imaging



Advances in


RoboticsJournal of










Advances in




ESMLDAMESVMKM0

20

40

60

80

100

Accu

racy

()

(g) Technology









40

50

60

70

80

90

100Av

erag

e acc

urac

y (

)


ESDBMPEMLE


40

50

60

70

80

90

100

Aver

age a

ccur

acy

()


ESDBMPEMLE


6040 1000 20 800

20

40

60

80

100

Aver

age a

ccur

acy

()


(c)





8 Conclusion



URMTCM

TEENSPFISL0

10

20

30

40

50

60

70Av

erag

e acc

urac

y (

)


URMTCM

TEENSPFISL0

10

20

30

40

50

60

70

Aver

age a

ccur

acy

()


6040 1000 20 800

20

40

60

80

100

Aver

age a

ccur

acy

()


(c)





Acknowledgments


References














































Advances in

FuzzySystems


Volume 2014












Journal of

Journal of





Advances in

Multimedia


Biomedical Imaging



Advances in


RoboticsJournal of










Advances in




40

50

60

70

80

90

100Av

erag

e acc

urac

y (

)


ESDBMPEMLE


40

50

60

70

80

90

100

Aver

age a

ccur

acy

()


ESDBMPEMLE


6040 1000 20 800

20

40

60

80

100

Aver

age a

ccur

acy

()


(c)





8 Conclusion



URMTCM

TEENSPFISL0

10

20

30

40

50

60

70Av

erag

e acc

urac

y (

)


URMTCM

TEENSPFISL0

10

20

30

40

50

60

70

Aver

age a

ccur

acy

()


6040 1000 20 800

20

40

60

80

100

Aver

age a

ccur

acy

()


(c)





Acknowledgments


References














































Advances in

FuzzySystems


Volume 2014












Journal of

Journal of





Advances in

Multimedia


Biomedical Imaging



Advances in


RoboticsJournal of










Advances in




URMTCM

TEENSPFISL0

10

20

30

40

50

60

70Av

erag

e acc

urac

y (

)


URMTCM

TEENSPFISL0

10

20

30

40

50

60

70

Aver

age a

ccur

acy

()


6040 1000 20 800

20

40

60

80

100

Aver

age a

ccur

acy

()


(c)





Acknowledgments


References














































Advances in

FuzzySystems


Volume 2014












Journal of

Journal of





Advances in

Multimedia


Biomedical Imaging



Advances in


RoboticsJournal of










Advances in














































Advances in

FuzzySystems


Volume 2014












Journal of

Journal of





Advances in

Multimedia


Biomedical Imaging



Advances in


RoboticsJournal of










Advances in












Advances in

FuzzySystems


Volume 2014












Journal of

Journal of





Advances in

Multimedia


Biomedical Imaging



Advances in


RoboticsJournal of










Advances in










Advances in

FuzzySystems


Volume 2014












Journal of

Journal of





Advances in

Multimedia


Biomedical Imaging



Advances in


RoboticsJournal of










Advances in



Documents

Research Article A Sentiment Delivering Estimate Scheme