AnOntologytoaidtheGoal-oriented …in1020/docs/seminarios/Monique... · 2017-09-13 · AnOntologytoaidtheGoal-oriented RequirementsElicitationandSpeciﬁcationfor Self-AdaptiveSystems

An Ontology to aid the Goal-orientedRequirements Elicitation and Specification for

Self-Adaptive Systems

Monique Conceição Soares

Ph.D. Thesis

Jaelson Castro (Advisor)Carla Silva (Co-Advisor)

Monique Conceição Soares CIn/UFPEAn Ontology to aid the Goal-oriented Requirements Elicitation and Specification for Self-Adaptive Systems 1 / 81

Outline

1 Introduction

2 Background

3 Analysis of the State of the Art

4 Onto4SASGORE Ontology

5 Onto4SASGORE Process

6 Evaluation

7 Conclusion


Outline

1 Introduction

2 Background




6 Evaluation

7 Conclusion


Introduction

Context- Self-adaptive Systems

Self-adaptivity is the ability of the system to adjust its ownbehavior in response to the environment. (BRUN et al., 2009).

- Requirements EngineeringElicitation;Identification;Documentation.


Introduction

Context- Requirements Engineering for SAS

Different from traditional approaches, RE for SASs focusesmore on defining adaptation logic.During RE for SASs, engineers must address what changes inthe environment and in the system to be monitored, what toadapt, when to adapt and how to adapt (YANG et al., 2014).


Introduction

Motivation

The goal-oriented modeling languages such as Tropos4AS(MORANDINI et al., 2008), AdaptiveRML (QURESHI et al., 2012) and DesignGoal Model (PIMENTEL et al., 2014) allow the goal-orientedspecification of the requirements for SAS, but they do notpresent any guide to elicit these requirements.


Introduction

Motivation

These languages were specifically created to model GORErequirements for SAS;They do not allow the specification of some key concepts forSAS:

modeling dimensions (ANDERSSON et al., 2009)

adaptation cycle (KEPHART et al., 2003);It can generate an insufficient specification.


Introduction

MotivationOntologies can be used on RE processes to reduce the negativeeffects of (CASTAÑEDA et al., 2010):

Insufficient specifications;Requirements not completely defined;The dynamic and changing requirements.


Introduction

Ontology is a standard way to represent concepts andrelationships on a specific domain (SOUAG, 2015).An ontology can be used and shared within a specific domainand it has been used to support requirements engineeringactivities (DERMEVAL, 2015).


Introduction

Motivation

The work presented by Qureshi et al. (2011) proposes a coreontology for SAS.It presents the context and resource concepts but it does notpresent all the key concepts related to the requirementsspecification for SAS.

modeling dimensions proposed by Andersson et al. (2009);an adaptation cycle (KEPHART et al., 2003).


Introduction

Core ontologyis a mid-term ontology that is not as specific as a domainontology but also not so domain-independent as a foundationalontology (DUARTE et al., 2016).

Foundational ontologyis a meta-ontology describing a set of real-world categoriesallowing to be used to talk about reality (GUIZZARDI, 2007).

Domain ontologiesare developed based on foundational ontologies. Referenceand Operational ontologies are domain ontologies (FALBO, 2014).


Introduction

Problem Statement

There is a need for a more complete ontology for SAS.It is necessary to define a domain ontology with concepts:

modeling dimensionsadaptation cycle

It is necessary to propose a domain ontology with all conceptsto represent:

contextmodeling dimensionsadaptation cycle

To specify more complete GORE requirements for SAS.It also is necessary propose a process to use the ontology.


Introduction

Objectives

Research objectiveDefine an ontology to support the requirements elicitation andspecification activities for SAS.

Research questionWhat are the essential concepts and relationships required tospecify GORE requirements for self-adaptive systems?


Introduction

Objectives

Research objectiveDefine an ontology to support the requirements elicitation andspecification activities for SAS.

Research questionWhat are the essential concepts and relationships required tospecify GORE requirements for self-adaptive systems?


Introduction

Specific Objectives:

1 Analyze the existing ontologies for SAS;2 Analyze the existing ontologies for context-awareness systems;3 Systematically organize the knowledge of SAS by analyzing

some GORE requirements languages proposed to specify SAS;4 Define a core ontology by extending the core ontology for SAS

(QURESHI et al., 2011), adding the change, mechanism, effect andadaptation cycle concepts and expanding the context concept;

5 Define a domain ontology for SAS, with concepts,relationships, axioms and competence questions;

6 Define a process to use the ontology with a GORE elicitationguide and a specification template;

7 Evaluate the ontology and the process.


Introduction

Research Methodology

Step 1: Analyzing the existing goal modeling languages forself-adaptive systems;Step 2: Performing a systematic literature review aboutontologies and SAS.Step 3: Performing a systematic literature review aboutontologies and Context-Aware Systems (CAS).Step 4: Building the core ontology.Step 5: Building a domain ontology.Step 6: Defining a process.Step 7: Evaluation. We used seven criteria to evaluate ourwork: completeness, verification, validation, usability,usefulness, ease of use and correctness.


Outline

1 Introduction

2 Background




6 Evaluation

7 Conclusion


Background

Self-adaptive SystemsOntology for context by Jeong-Dong et al., 2012.

Action, Goal, Rule, Time, Location, Status.MAPE-K from Kephart et al., 2003.

Monitor, Analyze, Plan, Execute, Sensor, Effector.Modeling Dimensions by Andersson et al, 2009.

Goal, Change, Mechanisms, Effects.


Background

OntologyMethodologies to create ontologies:

METHONTOLOGY (FERNANDEZ-LOPEZ, 1997).Uschold and Gruninger, 1996.SABiO (FALBO, 2014).


Outline

1 Introduction

2 Background




6 Evaluation

7 Conclusion


Analysis of the State of the Art

Ontology for Self-adaptive Systems: A SLR

We have performed two SLR;For the SAS domain we selected 13 works;For the CAS domain we selected 10 works;Totalizing 23 works analyzed as related works.


Analysis of the State of the ArtOntology for Self-adaptive Systems: A SLRThe SLR answered the following questions:

RQ01: Which groups and organizations are most active inontology for SAS research?RQ02: Which languages are used to describe the ontology?RQ03: Which research problems are addressed by theontology?RQ04: Which requirements engineering activities areconsidered by the ontology?RQ05: Which application domains are involved?RQ06: Does the analyzed ontology present just core concepts?RQ07: Does the ontology have a tool to support REactivities?RQ08: Which concepts are presented in the ontology?



Ontology for Self-adaptive Systems: A SLRRQ08: Which concepts are presented in the ontology?

Concept Cited byContext S4, S5, S6, S7, S8, S12Goal S1, S4, S6, S12Task S1, S5, S6, S12Agent S2, S3, S5, S8Change S9, S12, S13Sensor S7, S9, S11User S1, S7Actuator S7, S11

Some studies do not have concepts that clearly expresschange or adaptation, such as S1, S2, S3, S4 and S5 studies.



Ontologies for Context-Aware Systems: A SLRThe SLR answered the following questions:

RQ01: Which concepts are presented in the ontology?RQ02: Which languages are used to describe the ontology?RQ03: What application domains are involved in the paper?


Analysis of the State of the ArtOntologies for Context-Aware Systems: A SLRRQ01: Which concepts are presented in the ontology?The ten most cited concepts.

Concept Cited byLocation S14, S15, S16, S17, S18, S21, S22, S23Time S14, S16, S18, S19, S21, S22, S23Activity S17, S18, S19, S20, S21Role S18, S20, S21, S22, S23Environment S14, S18, S19, S21Network S14, S16, S17, S18Person S15, S17, S19, S20User S14, S15, S18, S21Agent S15, S17, S22Task S20, S21, S22


Analysis of the State of the ArtDiscussion of the SLRs

There are few works concerning ontologies for SAS.The limitation of these works in representing concepts for aSAS motivated the creation of a more complete ontology.The core ontology by Qureshi et al. (2011) is the S6 study.

Figure: Core Ontology for SAS by Qureshi et al. (2011).



Analysis of Requirements Modeling Languages for SASA self-adaptive news service example was modeled by:

Tropos4AS (MORANDINI, 2008);The Adaptive RML (QURESHI, 2012);The Design Goal Model (PIMENTEL, 2015).



Analysis of Requirements Modeling Languages for SAS

We noticed that these languages have several interestingconcepts, like Failure and Error (Tropos4AS), but they do notcover all the concepts necessary to specify the requirements ofSAS (SOARES et al., 2016) like the Mechanism concept.



GORE Languages versus Modeling DimensionsThe three languages present:

Goal, Softgoal, Plan, Task, Quality Constraint, DomainAssumption, Actor, Resource, Context, Failure, Error, Conflictand Condition concepts.

We note that the Condition concept can be interpreted as aQuality Constraint.



GORE Languages versus the Core Ontology for SAS

Each of the three modeling languages analyzed presentsdifferent elements from each other but they are unable tocapture some important characteristics of SAS.The core ontology for SAS (QURESHI et al., 2012) cannot represent allthe modeling dimensions neither the MAPE-K cycle or thecontext.


Outline

1 Introduction

2 Background




6 Evaluation

7 Conclusion


Onto4SASGORE: An Ontology for Self-adaptive SystemsGoal-Oriented REquirements

Onto4SASGORE organizes the knowledge of SAS andfacilitate the goal-oriented requirements elicitation andspecification.The Onto4SASGORE was built following three methodologies:

METHONTOLOGY (FERNANDEZ-LOPEZ et al., 1997);Uschold and Gruninger (1996);SABiO (FALBO, 2014).



We use the best practices of the three methodologies tocreate our ontology. The phases performed are:

Ontology Purpose Identification and Requirements Elicitation;knowledge acquisition;Conceptualization;Formalization;Implementation;Evaluation.



Purpose Identification and Requirements Elicitation:Objective;Scope;Motivating Scenarios;Informal Competence Questions;



knowledge acquisition:The roadmap papers presented in Cheng et al. (2009b) and inLemos et al. (2011);The modeling dimensions (ANDERSSON et al., 2009);The MAPE-K adaptation cycle (COMPUTING et al., 2006);The core ontology to performing RE for SAS (QURESHI et al., 2011);The CA5W1HOnto context ontology (JEONG-DONG et al., 2012);The three GORE modeling languages (Tropos4AS, AdaptiveRML and Design

Goal Model).The SLR’s results.



Conceptualization:Concepts, relations, properties and an OntoUML model;the CORE (JURETA et al., 2008), DOLCE and the UFO (GUIZZARDI, 2007)

as foundational ontologies;OntoUML and Menthor to create a graphical model.;



Figure: Core Onto4SASGORE in OntoUMLMonique Conceição Soares CIn/UFPEAn Ontology to aid the Goal-oriented Requirements Elicitation and Specification for Self-Adaptive Systems 36 / 81


Figure: Goal elements of the Core Onto4SASGOREMonique Conceição Soares CIn/UFPEAn Ontology to aid the Goal-oriented Requirements Elicitation and Specification for Self-Adaptive Systems 37 / 81


Figure: Context elements of the Core Onto4SASGOREMonique Conceição Soares CIn/UFPEAn Ontology to aid the Goal-oriented Requirements Elicitation and Specification for Self-Adaptive Systems 38 / 81


Figure: Modeling Dimensions elements of the Core Onto4SASGOREMonique Conceição Soares CIn/UFPEAn Ontology to aid the Goal-oriented Requirements Elicitation and Specification for Self-Adaptive Systems 39 / 81


Figure: MAPE elements of the Core Onto4SASGOREMonique Conceição Soares CIn/UFPEAn Ontology to aid the Goal-oriented Requirements Elicitation and Specification for Self-Adaptive Systems 40 / 81


Formalization:Create the axioms and formalize the competence questions andthe axioms;Competence Questions -> SQWRL;Axioms -> Description Logic.



Formalization:ID CQ01NL: What are the sensors in the scope of the system?FL: Sensor(?sen)->sqwrl:select(?sen)

ID CQ80NL: How many mechanisms are there in Change X?FL: hasMechanism(ChangeX, ?mec)->sqwrl:select(ChangeX) ∧ sqwrl: count(?mec)

ID CQ92NL: How does the system achieve the goals?FL: Goal(?goa) ∧ hasAction(?goa,?aci) ∧ makeSet (?s1, ?goa) ∧ makeSet (?s1,?aci) ∧

sqwrl:union(?s3, ?s1, ?s2) ∧ sqwrl:groupBy(?s3,?goa)->sqwrl:select (?s3)



Formalization:Formal axioms are assertions accepted as true aboutabstractions of a field (SOUAG et al., 2015).

Figure: Part of the axioms of Onto4SASGORE



Implementation:Protegé;SWRL;SQWRL;Manchester;Hermit.



Implementation:

Figure: Onto4SASGORE Metrics

Figure: Monitor ClassMonique Conceição Soares CIn/UFPEAn Ontology to aid the Goal-oriented Requirements Elicitation and Specification for Self-Adaptive Systems 45 / 81

Outline

1 Introduction

2 Background




6 Evaluation

7 Conclusion


Onto4SASGORE Process

How to use the ontology;Instance of the requirements elicitation and analysis processby Sommerville (2010);



Figure: Onto4SASGORE Process


Onto4SASGORE ProcessGoal-oriented Requirements Discovery Sub-process

Figure: Goal-oriented Requirements Discovery Sub-process



Goal-oriented Requirements Classification and Organization;Goal-oriented Requirements Prioritization and Negotiation;Goal-oriented Requirements Specification.


Onto4SASGORE ProcessGoal-oriented Requirements Specification

Figure: Onto4SASGORE Process



Goal-oriented Requirements SpecificationAction – <instance>Actor - <instance>Communicated Information - <instance>Goal - <instance>Context - <instance>Change - <instance>Effect - <instance>Mechanism - <instance>Resource - <instance>Sensor* - <instance>


Outline

1 Introduction

2 Background




6 Evaluation

7 Conclusion


Evaluation

Seven criteria:Completeness - by checking whether it presents the conceptsof the works selected the SLRs;Verification - The Onto4SASGORE can provide reliableanswers to competence questions using its terminology;Validity - it can be instantiated by some real world systems;Usability - we perform a case study with eight requirementsengineers, in which they use the Onto4SASGORE process;Usefulness and easiness of use - we performed a surveycreated through the TAM method;Correctness - Survey with specialists in SAS.


Completeness

With this criterion, we intend to prove that Onto4SASGORE ismore complete than the works selected in our SLRs.

Base of knowledge: Qureshi et al. (2011); the CA5W1HOnto(JEONG-DONG et al., 2012) ontology for context; the modelingdimensions for SAS (ANDERSSON, 2009) and the MAPE-K (KEPHART et

al., 2003) adaptation cycle.The S6 study is the ontology by Qureshi et al. (2011).the Onto4SASGORE is complete, considering that it canrepresent through its concepts all concepts of those works.


Verification

We answered the competence questions using theOnto4SASGORE elements;CQ01: What are the sensors in the scope of the system?SQWRL: Sensor(?sen)->sqwrl:select(?sen)

Answer: A55.CQ80: How many mechanisms are there in the Change X?SQWRL: hasMechanism(ChangeX, ?mec)->sqwrl:select(ChangeX) ∧ sqwrl:count(?mec)

Answer: Change. hasMechanism.CQ92: How does the system achieve the goals?SQWRL: Goal(?goa)∧ hasAction(?goa,?aci)∧ sqwrl: makeSet (?s1, ?goa)∧ sqwrl: makeSet

(?s1,?aci)∧ sqwrl:union(?s3, ?s1, ?s2)∧ sqwrl:groupBy(?s3,?goa)->sqwrl:select(?s3)

Answer: A13.


Validation

We instantiate the ontology with individuals extracted fromthe cleaner robot example;Protegé;Inferences through the Hermit reasoner.


Validation

Concept Instance from the Cleaner RobotActor Cleaner_robotChange Clean_at_programed_time

(hasSource: External;hasAnticipation: Foreseen;hasMechanism: Time_clock;hasChangeType: FunctionalType;hasFrequency: Rare;hasEffect: Clean_environment.)

NonFunctional_Goal Ensure_full_cleaning_coverage(hasFlexibility: Constrained)

Monitor Find_dock(hasOutput: Dock_location;before: Dock);Find_infrared_signal(hasOutput: Base_location;before: Way_to_base).


Validation

Concept InferenceActor The Cleaner_robot is also a SW_Agent.Change The Clean_at_programmed_time has duration

(hasDuration)Temporary and has evolution(hasEvolution) Dynamic.

NonFunctional_Goal The Ensure_full_cleaning_coverage is also aQuality_Constraint.

Monitor The Find_dock comes before To_Dock andCharge;The Find_infrared_signal comes before Get_on_baseand Return_to_base.


Usability

Eight requirements engineering answered an online applicationand they use the Onto4SASGORE process;The self-adaptive system used was a dispatch ambulancesystem;The collect data method is a survey.


Usefulness and Easy of Use

The survey was reasoned on Technology Acceptance Model(TAM) (DAVIS, 1989).Affirmatives about Usefulness.Affirmatives about Easy of Use.



About the Onto4SASGORE ontology perceived usefulnessAffirmative 1: The Onto4SASGORE ontology helps inelicitating requirements for SAS domain. (62.5%)Affirmative 2: The Onto4SASGORE is useful to guide therequirements elicitation for SAS. (75%)Affirmative 3: The Onto4SASGORE is useful in specifyingrequirements for SAS. (75%)Affirmative 4: I have difficulties in eliciting requirements forSAS. (50%)Affirmative 5: I have difficulties in specifying requirements forSAS. (50%)



About the Onto4SASGORE process ease of useAffirmative 6: The Onto4SASGORE process provides clearand sufficient information for eliciting requirements for SAS.(37.5%)Affirmative 7: The Onto4SASGORE process provides clearand sufficient information for specifying requirements for SAS.(62.5%)Affirmative 8: The description of each activity of theOnto4SASGORE process is clear and easy to understand.(62.5%)Affirmative 9: Training is required to use the Onto4SASGOREprocess. (87.5%)Affirmative 10: The Onto4SASGORE process is easy to apply.requirements for SAS. (25%)



The Onto4SASGORE is usefulness but the Onto4SASGOREprocess is not so easy of use.


Correctness

Survey with SAS specialists;Authors of the twenty three papers found in the two SLR;Research group at UFPE;Five specialists answered the survey.


Correctness

Questions:Question 01: Did you miss the representation of somehardware feature? What feature(s)?Question 02: Did you miss any adaptation information? Whatinformation(s)?Question 03: Did you miss the representation of somecontext-related information? What information(s)?Question 04: Did you miss any concept? What concept(s)?Question 05: Do you think the Onto4SASGORE ontology hasredundant elements? Which are them?Question 06: Do you disagree with the description of someconcept of the ontology? What concept(s)?


Correctness

Affirmatives:Affirmative 01: The Onto4SASGORE has the main conceptsfor SAS representation. (100%)Affirmative 02: The Onto4SASGORE ontology helps infinding new requirements. (80%)Affirmative 03: The Onto4SASGORE ontology helps inbuilding SAS systems. (80%)Affirmative 04: It is possible to elicit all the informationnecessary to specify a SAS by using the concepts presented inthis ontology. (100%)


Correctness

Affirmatives:Affirmative 05: The Onto4SASGORE has concepts to captureall information about SAS hardware. (60%)Affirmative 06: The Onto4SASGORE has concepts to captureall context-related information. (100%)Affirmative 07: The Onto4SASGORE has concepts to captureall information related to adaptation. (100%)


Correctness

The ontology was modified from the specialists answers inopen questions.The new version is presented in this thesis.


Outline

1 Introduction

2 Background




6 Evaluation

7 Conclusion


Conclusion

This thesis presents a work to define and create an ontologyto aid the GORE requirements elicitation and specificationactivities when developing self-adaptive systems.


Conclusion

Three approaches proposed for modeling goal-orientedrequirements to SAS were analyzed.

Tropos4AS (MORANDINI; PENSERINI; PERINI, 2008);Design Goal Model ((PIMENTEL, 2015);AdaptiveRML ((QURESHI; JURETA; PERINI, 2012).


Conclusion

Two systematic literature reviews were also performed toknow the works in the literature related to ontologies forself-adaptive systems and context-aware systems.


Conclusion

The Onto4SASGORE ontology has concepts related togoal-oriented requirements and SAS domains.Three methodologies to create ontologies were used:

Uschold and Gruninger (1996);SABiO (FALBO, 2014);METHONTOLOGY (FERNANDEZLOPEZ et al, 1997).

A process based on the requirements elicitation and analysisprocess by Sommerville (2010) was defined to guide its use.


Conclusion

To evaluate the Onto4SASGORE ontology and the process,seven criteria (completeness, verification, validity, usefulness,ease of use, usability and correctness) were used.

The results of this evaluation have shown that the proposedontology is complete, useful, usable, correct as well as it isverified and valid.


Conclusion

To evaluate the Onto4SASGORE ontology and the process,seven criteria (completeness, verification, validity, usefulness,ease of use, usability and correctness) were used.The results of this evaluation have shown that the proposedontology is complete, useful, usable, correct as well as it isverified and valid.


Conclusion

Contributions

Two systematic literature reviews with results of 23 studiesabout ontologies for SAS and CAS domains.

The analysis of three GORE modeling languages for SAS.An ontology to aid a requirements engineer to elicit andspecify GORE requirements for SAS.


Conclusion

Contributions

Two systematic literature reviews with results of 23 studiesabout ontologies for SAS and CAS domains.The analysis of three GORE modeling languages for SAS.

An ontology to aid a requirements engineer to elicit andspecify GORE requirements for SAS.


Conclusion

Contributions

Two systematic literature reviews with results of 23 studiesabout ontologies for SAS and CAS domains.The analysis of three GORE modeling languages for SAS.An ontology to aid a requirements engineer to elicit andspecify GORE requirements for SAS.


Conclusion

ContributionsThe competence questions

Identification of the concepts and relationships of the ontology;Evaluating of the verification and validation criteria.

The axioms were created to allow inferences in the ontology.The axioms were validated in Protegé.A process to help the requirements engineer to use theontology.The evaluation of the Onto4SASGORE through seven criteria.


Conclusion



The axioms were created to allow inferences in the ontology.The axioms were validated in Protegé.

A process to help the requirements engineer to use theontology.The evaluation of the Onto4SASGORE through seven criteria.


Conclusion



The axioms were created to allow inferences in the ontology.The axioms were validated in Protegé.A process to help the requirements engineer to use theontology.

The evaluation of the Onto4SASGORE through seven criteria.


Conclusion



The axioms were created to allow inferences in the ontology.The axioms were validated in Protegé.A process to help the requirements engineer to use theontology.The evaluation of the Onto4SASGORE through seven criteria.


Conclusion

Limitations

Some relevant work could be out of the SLRs;

The focus just in GORE modeling languages can be aconstraint to the concepts found;The sample of the survey with specialists can represent alimitation, because few specialists answered the survey.


Conclusion

Limitations

Some relevant work could be out of the SLRs;The focus just in GORE modeling languages can be aconstraint to the concepts found;

The sample of the survey with specialists can represent alimitation, because few specialists answered the survey.


Conclusion

Limitations

Some relevant work could be out of the SLRs;The focus just in GORE modeling languages can be aconstraint to the concepts found;The sample of the survey with specialists can represent alimitation, because few specialists answered the survey.


Conclusion

Limitations

Through the ease of use criteria of evaluation, we can realizethat the Onto4SAGORE is not so easy to use.

This work not contemplate the goal oriented requirementsprioritization and negotiation, though we know the importanceof this activity for a requirements process.


Conclusion

Limitations

Through the ease of use criteria of evaluation, we can realizethat the Onto4SAGORE is not so easy to use.This work not contemplate the goal oriented requirementsprioritization and negotiation, though we know the importanceof this activity for a requirements process.


Conclusion

Future Work

To facilitate the use, we plan to build a tool to support theontology use;Analyze the other modeling languages for SAS in order toknow the elements that theses languages can represent;Apply the ontology and the process in the specification ofother SAS;Perform case studies with more specialists;Contemplate the goal oriented requirements prioritizationactivity in the Onto4SASGORE process.


An Ontology to aid the Goal-orientedRequirements Elicitation and Specification for

Self-Adaptive Systems

Monique Conceição Soares

Ph.D. Thesis

Jaelson Castro (Advisor)Carla Silva (Co-Advisor)


Documents

AnOntologytoaidtheGoal-oriented …in1020/docs/seminarios/Monique... · 2017-09-13 · AnOntologytoaidtheGoal-oriented RequirementsElicitationandSpeciﬁcationfor Self-AdaptiveSystems