Download pdf - Bibliographic Classification in the Digital Age: Current

BibliographicClassificationintheDigitalAge:CurrentTrendsandFutureDirections

AsimUllah,ShahKhusro,and

IrfanUllah

INFORMATIONTECHNOLOGYANDLIBRARIES|SEPTEMBER2017

48

ABSTRACT

BibliographicclassificationisamongthecoreactivitiesofLibrary&InformationSciencethatbringsorderandpropermanagementtotheholdingsofalibrary.Comparedtoprintedmedia,digitalcollectionspresentnumerouschallengesregardingtheirpreservation,curation,organizationandresourcediscovery&access.Therefore,truenativeperspectiveisneededtobeadoptedforbibliographicclassificationindigitalenvironments.Inthisresearcharticle,wehaveinvestigatedandreporteddifferentapproachestobibliographicclassificationofdigitalcollections.Thearticlealsocontributestwoevaluationframeworksthatevaluatetheexistingclassificationschemesandsystems.Thearticlepresentsabird’s-eyeviewforresearchersinreachingageneralizedandholisticapproachtowardsbibliographicclassificationresearch,wherenewresearchavenueshavebeenidentified.

INTRODUCTION

Classificationistheprimaryinstinctofhumanbeingsinarranging,understanding,andrelatingknowledgeartifacts.Bibliographicclassificationprovidesaframeworkforarrangingandorganizingknowledgeartifactspreservedintheformofbooks,magazines,newspapersandotherholdingstoexplorenewavenuesofknowledgemanagement.TodayseveralclassificationschemesareinuserangingfromconventionalclassificationschemesincludingLibraryofCongressClassification(LCC),DeweyDecimalClassification(DDC),ColonClassification(CC),andUniversalDecimalClassification(UDC)toclassificationfordigitalenvironmentsincludingAssociationforComputingMachinery(ACM)digitallibrary1,InstituteofElectricalandElectronicsEngineering(IEEE)digitallibrary2,andOnlineComputerLibraryCenter(OCLC)cooperativecatalogue3.

Besidesthedifficultiesthatlieindevisingaclassificationscheme(time-consumingandresource-consuming),itisrequiredthateithertheexistingschemesshouldberevisedandextendedoranewclassificationschemeshouldbedevised,whichcouldactasacommonplatformforrepresentingknowledgeartifactsbelongingtodifferentcontexts.Suchaclassificationschemeshouldalsoresolvethechallengesindigitalpreservationandcurationandsupporttheprecise

AsimUllah([email protected]),ShahKhusro([email protected]),andIrfanUllah([email protected])areresearchersattheDepartmentofComputerScience,UniversityofPeshawar,Peshawar,Pakistan. 1http://dl.acm.org/2http://ieeexplore.ieee.org/Xplore/home.jsp3https://www.oclc.org/

BIBLIOGRAPHICCLASSIFICATIONINTHEDIGITALAGE|ULLAH,KHUSRO,ANDULLAH|doi:10.6017/ital.v36i3.8930

49

Andaccuratesearchandretrievalofdigitalcollections.Thefirststep,inthisconnection,istoproperlyanalyzeandevaluatetheexistingbibliographicclassificationschemesandtodigouttheirstrengthsandlimitationsinclassifyingdigitalcollectionsaccuratelyandappropriately.Therefore,theobjectivesofthisresearcharticleinclude:

• To investigate and evaluate the available approaches to bibliographic classification fromtheperspectiveofdevisingaclassificationschemethatcanactasacommonplatformforclassifyinganytypeofdigitalcollection.

• Todevise evaluation frameworks that compares the availablebibliographic classificationschemesandapproaches.

• Topresent issues,challenges,andresearchopportunities instate-of-the-artbibliographicclassificationresearch.

Therestofthepaperisorganizedas:Section2presentsthecurrenttrendsintheclassificationofdigitalcollections.Section3presentstwoevaluationframeworksforcomparingandevaluatingtheexistingsolutions.Section4presentsresearchchallengesandopportunitiesinbibliographicclassificationresearch.Finally,Section5concludesourdiscussion.Referencesarepresentedattheendofthepaper.

ClassifyingDigitalCollections–AMixedTrend

Thebibliographicclassificationhasbeenthefocusofseveralresearcherstoproperlyclassify,catalogue,anddescribedigitalcollections.Inthisregard,twoapproacheshavebeenadopted:theformersupportstheuseofconventionalclassificationschemesincludingCC,DDC,andLCCetc.,indescribingandclassifyingdigitaldocuments,whilethelatterrecommendsdevisingsomenewwaysofclassificationsuchasACM4computingclassification.However,inmostofthedigitalenvironments,amixedtrendhasbeenobserved,wherealongwithnewclassificationschemes,categorizationisalsousedasacomplementarysolution.Forexample,ACMpresentsitsownclassificationsystemaspoly-hierarchicalontologyindescribingComputerScienceliteratureandforusinginSemanticWebapplications.ACMhasreplacedits2008ACMclassificationsystemthatservesasde-factomodelfortheclassificationofComputerScienceliteraturebygivingvisualtopicdisplayalongwithsearchingservices.Itservesassemanticvocabularyforcategorizingconceptsandafoundationofcomputingdisciplines("The2012ACMComputingClassificationSystem,").Similarly,IEEEdigitallibrarycategorizesitsholdingsintodirectoriesperitsownrulesofcataloguingandcategorization.Itcategorizesarticlesandstandardsintoseveralsubjectareasandclustersdocumentsthroughyearofpublication,authornames,contenttype,affiliation,publicationtitle,publisher,countryofpublication,alphabets,numeralsandalphanumericvalues5.Thedocumentcollectioncanbenavigatedthroughcollectionnames,numberofdocuments,bytopicandInternationalClassificationforStandards(ICS).

4http://dl.acm.org5http://ieeexplore.ieee.org/browse/standards/ics/ieee/


50

TheDMOZ6directoryisthelargesthumanmadedirectoryofwebpages.Sinceitsinceptionin1998,itcategorizes3,861,137websitesavailablein90languagesinto1,031,719categoriesandsub-categoriesby91,928editorsandvolunteers.Inaddition,ithasitsDMOZRDFdumpsavailableonLinkedOpenData(LOD)cloud.AccordingtotheWorldWideWebConsortium(W3C),LODenablesthedataintegrationandreasoningatalargescale("Linkeddata,").Itestablisheslinksamongdataenablingmachinesanduserstoexplorethewebofdataratherthanthewebofdocumentsalongwithfindingrelateddata(Berners-Lee,2006;Bizer,Heath,&Berners-Lee,2009).However,itlacksinsemanticsearch(meaningfulsearch),whichaffectstheprecisionandaccuracyinexploringtherequiredresources.Also,thecategories,underwhichthewebsitesarekept,areneededtoberevisedbecausetherecanbefacetedandintra-hierarchicallinksamongwebpages.Inaddition,thecontentmanagementneedstobeupgradedforupdatingthedirectorywithnewentriesandthewayitreviewsandcategorizeswebsites(Boykin,2016).

Institutionalrepositoriesusethemixedapproachtowardscreating,collectingandmanagingmetadataforprintedanddigitalcollectionsusingseveralsourcesincludingconventionalanddigital.Thismixedtrendintroduceschallengestothemetadatamanagers(Chapman,Reynolds,&Shreeves,2009).Todealwiththesechallenges,thesubjectclassificationsystemscanbeverybeneficialinprovidingWeb-orientedservicesincludingsearchingofcontentsthroughsearchpatterns,browsing,andcontentfilteringbysubjectarea.However,atthesametime,acognitiveoverloadrisesfortheauthorsanddepositorsoftheinstitutionalrepository(Cliff,2008)thatneedsfurtherattention.

Tohandletheinformationoverloadinretrievingdigitalcollections,severalcontrolledmethodshavebeenproposedintheliteraturerangingfrommanualtechniques(e.g.,webdirectories)toautomatictechniquesincludingclusteringandclassification.Severalclassificationschemesincludingsentimentandsubjectclassificationhavebeendevelopedforclassifying(andcategorizing)webpages.Classificationisusedinfocusedcrawling,searchingandrankingresults,andclassifyingqueries.Clusteringalsoclassifieswebresourcesbutitisslightlydifferentfromclassification,whichisbasedonarigidpredefinedtaxonomyandrulesforinterpretingthemeaningofclassificationorder.Ontheotherhand,clusteringshowsflexibilityinclassification(categorization)ofwebdocuments(Zhu,2011).However,amixedtrendhasbeenobserved,whereclassificationandcategorizationareintermingledtofacilitateorganization,description,exploration,andretrievalofdigitalcollections.

SemanticWebbringsmeaningfulconnectionsbetweenthewebofdatasothatnotonlyhumansbutmachinescanalsounderstandthecontentofdocumentstoretrievethemostintendedandrequireddocuments.Thiswayotherrelateddocumentscouldalsobeeasilyconnectedandretrieved(Berners-Lee,2006).Tounderstand,describe,andrelateconceptswithindocuments,ontologiesareused.Therefore,researchershavebeenworkingonbringingsemanticsthroughSemanticWebandrelatedtechnologiestoautomaticallyclassifydigitalcollections.Forexample,

6http://www.dmoz.org


51

(Beghtol,1986)arguesthatsemanticaxismakessyntacticalclassificationstructuremoremeaningfulandprovidestheplatformfordevelopingrelationshipsamongknowledgeartifactsthroughseveralwarrantsinclassificationsystems.Similarly,classificationontologyisusedinautomaticclassification(Wijewickrema&Gamage,2013)tominimizetheambiguityinvocabulary.Toobtainasinglesubjectfortheinputdocument,severalweightfunctionsincludingthetermfrequency-inversedocumentfrequency(TF-IDF),andfilteringmethodsareapplied.

SemanticWebandLODtechnologieshavealsobeenusedindealingwithbibliographicdata.Forexample,BibBase7,abibliographicdatapublishingandmanagementtool(Xin,Hassanzadeh,Fritz,Sohrabi,&Miller,2013)publishesbibliographicdataontheuserwebsiteaccordingtoLODprinciples.However,thesearelimitedbecauseofthelackofinteroperabilityamongnativelanguageswhiletranslatingclassificationrecordsfromsourcelanguagetothetargetlanguage(Kwaśnik&Rubin,2003).

Theclassificationschemesarealsobeingconvertedintoontologies.(Giunchiglia,Marchese,&Zaihrayeu,2007)haveappliedreasoningcapabilitiesofOWLontologiestoclassificationschemes.Theseontologiesareusedasinterfacestohumanknowledgeformachineswhereasclassificationschemesareinterfacestoknowledgeforhumans.However,thereislimitedsupportavailableforcross-disciplinarysearchingandaccommodationformoreviewsandinterpretationsofknowledge(Albrechtsen,2000).Thesupervisedandunsupervisedmachinelearningtechniquesareusedforautomatictextclassification.SupervisedmachinelearningtechniquesusemodelsincludingmultinomialNaïveBayesmodel,andBernoullimodel(Manning,Raghavan,&Schütze,2008)forclassification.Yelton(2011)appliesprobabilisticclassificationofimportantwords(andthereforeofdocuments)especiallybyconsideringAmazon’sStatisticallyImprobablePhrases(SIPs)8andGooglephrasesearchinsideabook.Forsubjectanalysis,hementionssimplistic;content-based;andrequirements-basedmethodsintermsofunderstandingtextclassificationandmanipulationofbooks.TheWikipediapagestructuralhierarchyisexploitedinautomaticallyharvesting,classification,categorization,clustering,andmetadataenrichment(Yelton,2011).

InformationExtraction(IE)isalsoappliedinclassifyingbooksautomatically.Forexample,(Betts,Milosavljevic,&Oberlander,2007)useIEmethodsforautomaticlabelingofbooksusingLCCclassification.Theyusedbag-of-words(BOW)model,bag-of-named-entityrecognition(NER)model,generalizingnamedentities(GAZ)modelinautomatictextclassification.Toachievebetteraccuracy,theyalsocombinedtheresultsofthesemodels.Howeverautomaticclassificationmayleadtolimitedsearchandretrievalbecauseofthemissingsemanticsassociatedwithphrasesorkeywords.Toovercomethisissue,afundamentalandpracticaltheoreticalmodelofclassificationisrequired(Jones,1970).

7https://bibbase.org/8http://www.amazon.com/gp/search-inside/sipshelp.html


52

Table1categorizesthebibliographicclassificationapproachesintothreebroadercategoriesnamely:theoreticalapproaches,practicalapproachesandapproachesusedindigitalenvironments.Theoreticallyresearchershavediscusseddifferentviewpointsforclassification,whereaswegetadifferentviewwhentheseschemesareappliedforclassification.Practically,thesyntacticstructureisvaluedbyusingfacetedandenumerativetechniques.IndigitalenvironmentsliketheWebanddigitallibraries,strictboundariesofclassificationareoftencompromisedbycategorization.

ApproachestoClassification

TechniquesUsed

TheoreticalApproaches

1. Biasness(Mai,2009)(Mai,2010)2. Subjectivityandobjectivity(Hjørland,2016)3. EpistemologicalandSemioticapproaches(Hjørland,2013)

(Lee,2012;Mai,2011)(Tennis,2008)4. Empiricism,Rationalism,HistoricismandPragmatism

(Hjørland,2013)5. Multidisciplinarityapproach(Beghtol,1998)6. Scientificapproaches(Hjørland,2008)7. Positivisticandpragmaticapproaches(Dousa,2009)(Mai,

2011)8. Interdisciplinaryandevidencebasedpracticeclassification

(Hjørland,2016)9. Socialandculturalcontext(J.-E.Mai,2004)10. Bytrackingtheuniverseofknowledge11. Universalorder(Smiraglia&VandenHeuvel,2011)12. Integrativelevelsinclassification(Dousa,2009)13. Literarywarrant(Rodriguez,1984)14. Educationwarrant(Hjørland,2007)(Beghtol,1986)15. Semanticwarrant(Beghtol,1986)16. Syntacticwarrant(Beghtol,1986)17. Domainandusersrequirements(Mai,2005)18. Pluralismandhumaninterpretations

PracticalApproaches

1. EnumerativeandFaceted(Batley,2014).2. GeneralPurposeapproach(Mai,2003)andSpecialPurpose

approach(Mancuso,1994)e.g.classificationschemesforgeneralclassesofknowledgeareasorforaspecialclassofknowledgearea.

3. Syntacticaxis(Beghtol,1986)(Beghtol,2001)4. Semanticaxis(Beghtol,1986)(Beghtol,2001)


53

ClassificationinDigitalEnvironment

1. DocumentSimilarity(HammingdistanceandEuclideangeometricapproaches)(Losee,1993)

2. Fuzzyapproach(Jacob,2004)3. Clustering(Nizamani,Memon,&Wiil,2011)4. Categorization(Koshman,1993)5. TF-IDFweighting(Dorjietal.,2011)6. Unsupervisedmachinelearningtechniques(Joorabchi&

Mahdi,2011).(K-meanClustering,hierarchicalclustering)7. Supervisedmachinelearningtechniques(Wang,2009)

(MultinomialNaïveBAYES,Bernoullimodel,SupportVectorMachine,RandomForest,K-NNtechnique)

8. InformationExtractionmethods(Gilchrist,2015)9. Probabilistictextanddocumentclassification(Maron,Kuhns,

&Ray,1959)10. Ontologies(Campbell,2002)

Table1.Categorizationofapproachestowardsbibliographicclassification

EvaluatingClassificationSchemes&Approaches

InthisSection,wepresenttwoevaluationframeworkstocompareandevaluatetheexistingclassificationandcategorizationsystemsandwell-knownbibliographicclassificationontologies.WehavechosenCC,DDC,LCC,andUniversalDecimalClassification(UDC)onthebasisoftheirstructuralpropertiesandwideusagebothinconventionalanddigitallibraries("Subjectclassificationschemes,"2015)("LibraryofCongressClassification,"2014)("AboutUniversalDecimalClassification(UDC),")(Press,2002)(Encyclopedia,1August2014).Someofthesepropertiesinclude:citationandfillingorder;notationsexpressiveness;flexibilityinclassificationprinciples,rulesandnotations;coverageoftheknowledgeareas;classificationschedulesandnotationsstructure;notationsbrevityandsimplicity;notationsmnemonics;notationshospitality;scheduleswithupdateableandcomprehensivesubjectsorder;andknowledgecoverage(Batley,2014).TheUDC,LCC,andDDCareuniversal,multidisciplinary,andwidelyusedsystems(Koch&Day,1997),whereasCChastheseminalandinspirationalvalueforthefacetedstructureofthebibliographicclassification.Therefore,theevaluationframeworkmainlytargetstheseclassificationschemesasournaturalchoicefortheevaluationandcomparison.Similarly,weevaluateACM9,IEEE10andDMOZ11usingtheevaluationframeworkasthesearethewell-knownandwidelyuseddocumentclassification&categorizationsystemsforthedigitallibraries.Table2presents22metricsusedintheevaluationframework.Theseevaluationmetricsareextracted

9http://www.acm.org/about/class10http://www.ieee.org/about/today/at_a_glance.html?utm_source=mm_link&utm_campaign=iaa&utm_medium=ab&utm_term=at%20a%20glance11https://www.dmoz.org/docs/en/about.html


54

fromtheexistingliterature(Kaosar,2008)(Painter,1974)(Encyclopedia,1August2014)(Buchanan,1979)(Kaosar,2008)(Painter,1974)(Encyclopedia,1August2014)(Kochetal.,1997)(Reiner,2008)(Gnoli,Merli,Pavan,Bernuzzi,&Priano,2008)(Francu,2007)(Chan,Intner,&Weihs,2016).Thesemetricsinclude:(i)structuralcomplexity;(ii)notationalbrevity;(iii)predefinedstructure;(iv)rulescomplexity;(v)theoreticallaws;(vi)mnemonics;(vii)hospitality;(viii)searchcomplexity;(ix)usability;(x)precisionandaccuracy;(xi)multilinguality;(xii)interoperability;(xiii)semanticsearch;(xiv)biasinsubjectrepresentation;(xv)enumerativestructure;(xvi)facetedstructure;(xvii)facetedsearch;(xviii)consistency;(xix)LODdatasets;(xx)LinkedOpenVocabularies(LOV)support;(xxi)platform;and(xxii)warrantsofclassification.Thesemetrics,theirneed,andtheiruseinratingsofclassificationsystemsarediscussedinthefollowingparagraphs.InTable2,thesebibliographicsystemsareevaluatedforthesemetrics.Theindicatorüshowsthepresenceofmetricvalue,ûindicatorrepresentsthatthesystemhasnoorminimalsupportforthementionedmetric,whereasandN/Aisusedfornotapplicable.Inaddition,eachclassificationsystemhasbeenevaluatedandratedbasedonthesemetrics(Table3),whereFigure1graphicallydemonstratestherankingsandratingsoftheseclassificationsystems.

SchemesMetrics

CC UDC DDC LCC ACM IEEE DMOZ

StructuralComplexity

ü ü û û û û û

NotationalBrevity

û û ü ü ü ü N/A

PredefinedStructure

ü û ü ü ü ü ü

RulesComplexity

ü ü û ü û û û

TheoreticalLaws

ü ü ü ü û û û

Mnemonics ü ü ü ü ü ü ûHospitality ü ü ü ü ü ü üSearchComplexity

ü ü û û û û ü

Usability ü ü ü ü ü ü ûAccuracyandPrecision

ü ü ü ü ü ü û

Multilinguality ü ü ü ü û û üInteroperability û ü ü ü ü ü ûSemanticsearch


Biasinrepresentation


EnumerativeStructure

û ü ü ü û û û


55

FacetedStructure

û ü û û ü ü û

FacetedSearch û ü ü ü ü ü ûConsistency ü ü ü ü ü ü üLODDatasets û ü ü ü ü ü üLOVSupport û û û û ü û ûPlatform N/A UDC

consortiumOCLC Library

ofCongress

ACMdigitallibrary

IEEEXploredigitallibrary

OpenDirectoryProject

Warrantsofclassification

LiteraryWarrant(Giess,Wild,&McMahon,2007)

LiteraryWarrant(Perles,1995)

LiteraryandScientificWarrant(Giessetal.,2007)

LiteraryandScientificWarrant(Giessetal.,2007)

ScientificResearchwarrant

ScientificResearchwarrant

N/A

Table2.EvaluationofClassificationSchemes

Thestructuralcomplexitymeansdifficultiesinusingthestructureandnotationsinclassifyinganddescribingaspecificsubjectarea.Themetricwillhelpusinselectingaclassificationschemeorsystemthatiseasytouseinclassifyingdocumentcollectionbyrequiringshortnotationsandsimplerules.ThenotationsandrulesarecomplexinCCandUDC(Ranganathan,1968).Thiscomplexityisbecauseofthefacetedstructureintheseclassificationschemes(Sukhmaneva,1970).ThestructuralcomplexityofCCisgreaterthanthatofUDC.UDCcomesatsecondpositionincomplexityascomparedtoCC.Becauseofitsenumerativestructure,LCCstandsatthirdposition,asitislessercomplexthanCCandUDC.DDCisthesimplestinthislistbecauseitisbasedonenumerativeclassificationstructureandontheprincipleofdividinguniverseofknowledgeintodefinedclasses.IEEEismorecomplexthanACM,whereasDMOZistheleastcomplexsystem.Theclassificationsystemwithgreaterstructuralcomplexityisrankedlowerinthelist.Therefore,basedonthismetric,theclassificationsystemscanberankedasDMOZ,ACM,IEEE,DDC,LCC,UDC,andCC.

Thenotationalbrevitymeanshowbriefarethenotationsindescribingandunderstandingtheholdingswithminimumnumberofsymbolsandminimalcognitiveload.DDCuseswell-organizedshortnotationsandtheirmnemonicvalueisalsogreater(Comaromi&Satija,1983)(Hyman,1980).LCChasnotationalbrevity(Chanetal.,2016).UDCuseslengthynotations(Kaosar,2008)ascomparedtoDDC,whereasCCalsouseslengthyandcomplexnotations(Chatterjee,2016).ACMnotationsareshorterthanIEEE,whereasDMOZdonotuseanynotationsatall.Usingthismetric,theseclassificationsystemscanberankedasACM,IEEE,DDC,LCC,UDC,CC,andDMOZatthelastwithnousageofsymbolsatall.

Thepredefinedstructuremeansthattheclassificationschemefollowsrigidpre-assumedsubjectcategorizationalongwithclassificationclassmarks.Inthisregard,UDCandLCCareenumerative


56

andimposesubjectivityviewpointofclassificationbyfollowingapredefinedstructure(Goh,Giess,McMahon,&Liu,2009).Beingfaceted,CCarrangesbasicconceptsinfewpredefinedcategories(Satija&Martínez-Ávila,2015).DDCalsohasthepredefinedhierarchicalstructureofclassification(Press,2002)(Jonassen,2004).Amongtheseschemes,CChasminimalpredefinedstructurebecauseofusingfacets;UDCisbothenumerativeandanalytico-synthetic.LCCisenumerativebutpossessesweakerpredefinedrulesforthestructuraldesign.Becauseoftherigidenumerativehierarchiesandpredefinedclassstructure,DDCcomesatfirstposition.DMOZhasthemostrigidpredefinedstructureascomparedtothatofIEEEandACM.Theclassificationsystemwithmostrigidandpredefinedstructureisrankedlower,andtherefore,therankingcouldbeCC,ACM,IEEE,UDC,DDC,LCCandDMOZ.

Thecomplexityinrulesdeterminesthedifficultylevelinapplyingclassificationrulesonknowledgeartifacts.CCpresentsacomplexsetofrulesandclassificationtheory,whichiscomparativelydifficulttoimplementandunderstand(Tennis,2011).LCCisalsocomplex("LibraryofCongressSubjectHeadings:Pre-vs.Post-CoordinationandRelatedIssues,"March15,2007)inimplementingLibraryofCongressSubjectHeadings(LCSH)inpre-coordinatedsubjectstrings.DDC’srulesandprinciplesarecomprehensiveandcomplete(Press,2002)andeasierthanthoseofCCandLCC.UDCisalsoeasytounderstandandimplement(Piros,2014).ACM,IEEE,andDMOZaresimpletouseandunderstand,andtherefore,bearsnosuchcomplexity.Aclassificationsystemwithgreatercomplexityisrankedlower,therefore,basedonthismetric,therankingscouldbeACM,IEEE,DMOZareontopwithsimilarrankingsfollowedbyUDC,DDC,LCC,andCC.

Theoreticallawsareconsideredasametrictoanalyzethefoundationsofclassificationsystemstounderstandwhethertheyarebasedoncertaintheoreticallawsandprinciplesofclassificationornot.UDCcombinestheenumerativeandfacetedapproachesgatheredfromDDCandCC(Kaosar,2008).ThesyntheticprincipleofUDCcontributestoitswidespreadusebutitisnotenoughattheintellectuallevelformakingtherelationsbetweenthesubjectfacets(Kyle&Vickery,1961).UDClacksstandardrulesforitsapplicationformakingfacets,buttherearerulesforitsstructuralrepresentation(McIlwaine,1997).Therefore,thestructuralandsyntheticrulesaregoodenoughforitsapplicabilitybutitshouldberefinedfurtherattheintellectuallevel.ThetheoreticallawsofCCarebasedonthefacetedapproachofmanagingknowledgeartifacts.CChassoundrulesandprinciples,whichincludedifferentpostulates,laws,principlesandcanons(Batley,2014)(ArashanapalaiNeelameghan&Parthasarathy,1997).Ontheotherhand,LCChasweakertheoreticalfoundations.Therealsoexistssomeintellectualandstructurallimitationsduetoitsenumerativestructure(SanSegundoManuel,2008).DDChasthehierarchicalandtheenumerativestructurewhichisbasedontheknowledgephilosophyofhierarchicaldivision(Hjorland,1999).Becauseofthestrongtheoreticalfoundations,CCisatthetopofthislist,DDCissecondbecauseofitsuniversaltheoryofknowledgedivision,UDCisthirdforbeingexploitingthetheoriesofDDCandCC,LCCisatfourthpositionforcomparativelyweaktheoryofclassification,whereasACM,IEEE,andDMOZpresentnoorverylimitedtheoreticallawsorphilosophicalrulesofclassification.


57

Thesupportforusingmnemonicsenableshumanclassifierstoeasilymemorizethesymbolsandnotationsofclassificationscheme.ThesystematicandliteralmnemonicsareusedinUDC(Satija,2013)(Kaosar,2008).Themnemonicsareincreasedthroughmnemonicdevices,whicharedescribedthroughthecanonsofmnemonics(Kaula,1965).LCCusesliteralmnemonics(Satija,2013),whereasDDCusessystematicandliteralmnemonicsbutitssystematicmnemonicsarenotconsistent(Satija,2013).ThereareseveralseminalmnemonicsinCC(Rahman&Ranganathan,1962).ThesemnemonicdevicesincreasemnemonicsinCC,buttheformationandlengthofthenotationsaffectsthismnemonicquality.ACMhasgreatersupportformnemonicsincomparisonwithIEEE,whereasDMOZisthecollectionofwebpagesunderspecificcategories.Basedonthismetric,therankingsofclassificationsystemscouldbeDDC,UDC,LCC,ACM,IEEE,CC,whereasDMOZlacksinusinganymnemonicdevicesornotations.

Hospitalitymeanstheabilityofaclassificationschemetoincorporatenewknowledgeareasexpressedindifferentmultilingualcontexts.HospitalityispresentinUDC(Kaosar,2008).CCisalsohospitablefornewsubjects(DeGrolier,1962).LCCishospitableforexpressingthenewsubjectsandknowledgeareas(Satija,2013).DDCishospitablefornewsubjectareas(Satija,2013).Byapplyingthismetric,aclassificationschemewithfacetedapproachisnaturallymorehospitablethanothers.Therefore,CCismorehospitableandatthetopinthislistfollowedbyUDC.DDCisatthirdpositionforbeingfollowingenumerativeapproach.LCCisatfourthpositionbecauseofit’sofpureenumerativestructure.IEEEandACMareatfifthpositionbycoveringshortspanofknowledgeareas,facetedstructure,andefficientsearch.DMOZiscoveringonlywebpagesinalreadyspecifiedcategoriesthereforeitisatseventhposition.

Searchcomplexitymeasuresthedifficultyinsearchingartifactsusingaclassificationscheme.Itdescribesthatwhichclassificationschemeisworthinsearchingaspecificdocument.SearchcomplexityisminimalinUDCbecauseofitssyntactic-analyticoandenumerativenature(Kaosar,2008),whichcancontributeinsearchapplicationsinbothWebbasedandin-housesearchingapplicationse.g.,OnlinePublicAccessCatalog(OPAC).ThetheoryandphilosophyofCCisthetrendsetterfortheknowledgemanagement,resourcediscovery&access,however,accordingto(Raghavan,2016)searchingthroughCCiscomparativelyweakerthanotherbibliographicclassificationschemes.Accordingto(Chan,2000),LCCandLCSHhavethepotentialtoprovidetheeaseinsearchingbecauseofrichervocabularyforgreatersubjectcoverage,synonymandhomographcapabilities,pre-coordinatedsystem,browsingcapabilityinmulti-facetedstructure,multilingualsupportandMARCformatsupportwithsematicinteroperability.However,itislimitedinprovidingeaseinsearch&retrievalprocess,whichincludesyntaxandapplicationrulescomplexity,lackoftrainingforthepersonnel,andtoolengthyandcomplexsearchingstrings.DDCandLCCareaggregatedinClassify12projectinitiatedbyOCLC.WiththeuseoftheClassifyapplication,thesearchexperienceofthecatalogersandpatronsbecomesmucheasier.Usingthismetric,DDCstandsatthetopwithleastcomplexitythanLCC,UDC,andCC.IEEEismorecomplexthanACMandDMOZ.Theclassificationschemewithlesssearchcomplexitywillberankedhigher. 12http://www.oclc.org/research/themes/data-science/classify.html


58

Therefore,ACMandIEEE,DDC,andLCCstandfirstwithleastsearchcomplexityfollowedbyUDC,andCC.DMOZstandsatthelastpositionwithgreatersearchcomplexityhavinglooseboundariesofcategorization.

Usabilityanalyzesthedifficultyinusingaclassificationschemeforclassifyingandsearchingdocuments.Thismetricdefinestheeaseoflearningandeffectiveusage.Usabilitymeasuresusersatisfaction,userunderstandingofthesystem,andprecisionwithminimalrecallinlesseramountoftime(Singapore,2016).OCLChasincludedstructuralchangestoimproveusabilityandsimplifyclassificationtasks("DeweyServices:DeweyDecimalClassificationSystem,").TheClassify13projectaimsatfindingbooksthroughawebinterface,whichiseasytouseandunderstandbyusingDDCandLCC.UDCisextensivelyusedinWeb-basedsearchandretrievalapplications(Kaosar,2008).Thisclassificationschemeisusedinseveralinstitutions’OPACsystems("LibraryOPACscontainingUDCcodes,").TheUDCnotationsaresupportivefortheusability(Slavic-Overfield,2005).However,theuserinterfaceoftheseOPACsearchsystemscouldbefurtherimproved(Slavic,2006)(Pollitt,1998)(Schallier,2005).CCisthesourceofinspirationandastandardizedmodelfortheusabilityoffacetedstructureofbibliographicclassificationintheelectronicandwebbasedenvironments(Thelwall,2009).In(Rosenfeld&Morville,2002),thephilosophyandmethodologyofCCisconsideredattheabstractandtheoreticallevel.ThisassessmentofCCleadsustotheargumentthatthefacetedstructureissupportiveinpreciseretrievalwithaconsiderablyhighcognitiveworkattheuserendascomparedtoDDCandLCCbecauseoftheirsimpleenumerativestructures.LibraryofCongressusesLCCinitscatalog14andClassificationWeb15applications.TheseapplicationsareexploitingLCSHsandLCCinuserfriendlymanner.Bylookingattheusabilityaspectoftheseclassificationschemes,therankingthroughthismetricsappearsasDDCisatthetopforitseasyenumerativestructureandnotationalsimplicityalongwitheasytouseWebapplications.LCCisatsecondpositionbecauseofitsenumerativestructureandadoptabilityinwebapplications.Beingenumerativeandfaceted,UDCstandsatthethirdposition.CCforbeingapurefacetedschemewithcomplexnotationsandrules,isrankedatthefourthposition.Similarly,IEEEandACMarefacetedandeasytouse,andtherefore,sharethefirstpositionwithDDC.DMOZwithlooseboundariesofcategorizationisleastusablewithlimitedbrowsingandsearch.

Theaccuracyandprecisionmetricmeasureshowaccurateandpreciseaclassificationsystemcanidentifytheexactlocationsoftheholdingsinthegivenknowledgespace.UDCshowsaccuracyandprecisioninfindingtherequiredknowledgeartifact(Kaosar,2008).TheaccuracyandprecisionofCCgetscompromisedasitslengthynotationsintroducescomplexityinsearchinganddiscoveringdocuments(Satija,2015).LCCandDDCwereresearchedforaccuracyandprecisionbyusingaprototypemodel(Gnoli,Pusterla,Bendiscioli,&Recinella,2016)forautomatictextclassificationofelectronicdocumentsusingclassificationmetadataoflibraryholdingsfromLCCandDDC

13http://classify.oclc.org/classify2/14https://catalog.loc.gov/vwebv/searchBasic15https://www.loc.gov/cds/classweb/classwebfeatures.html


59

datasets.Itwasobservedthatforprecision,thereisaneedforincreasingDDCandLCCbibliographicdataontheWeb,introducingsearchingcapabilitiesforbibliographicdataatthemicrolevelofanydocument,andincreasingtheefficiencyofuserinterfacesfornavigationusingDDC-basedbrowsingstructure(Joorabchi&Mahdi,2009)(Joorabchi&Mahdi,2011).Therefore,CCbecauseofthepurefacetedapproachhashigh-levelprecisioninsearchandresourcediscovery.UDCstandssecondbecauseofbeingenumerativeandenumerativeandanalytico-synthetic.DDCisatthethirdpositionasOCLCmaintainsandupdatesitsstructureregularlyalongwithstate-of-the-artsearchapplications.LCCsharesthethirdpositionwithDDC,beingregularlyupdatedandmaintainedbyLibraryofCongressforprecisionintheirsearchapplication.IEEEandACMalsoshowgreaterprecisionintheirsearch&retrieval,andtherefore,sharethethirdpositionwithDDCandLCC.DMOZarethemanuallycreatedandupdatedcategoriesofwebpages,havinglimitedkeywordsearchwithverylowprecision.

Inconnectiontotheevaluationframework,multilingualitymeanstoclassifyanddescribetheknowledgeartifactswrittenandexpressedinvarietyofnaturallanguagesandtheavailabilityofanyclassificationschemeindifferentnaturallanguages.DMOZsupports72differentlanguagesoftheworldandthereforestaysatthetop.UDCismultilingualbysupportingFrench,Portuguese,SpanishandRussian(Slavic,2008)(Koch&Day,1997)andhasbeentranslatedintolanguages("UniversalDecimalClassificationsummary,"2017).LCCsupportsworksin19languagesubclasses("LibraryofCongressClassificationOutline:ClassP-LanguageandLiterature,")includingGerman,Slavic,OrientalLanguagesandRomanlanguagesetc.ThetranslationsofDDCsupporttolocalizethisschemefordifferentlanguagesoftheworld(Vizine-Goetz,2009).DDCistranslatedin30differentlanguagesbutcoversdifferentlanguagesinonlysevenclassesi.e.,from420to490classnumber("DeweyDecimalClassificationsummaries,").CCshowsminimalmultilingualsupportbecauseofitssub-continentalorigin(ANeelameghan&Lalitha,2013;Raghavan,2016).ACMandIEEEareinEnglishlanguagesonlyandtherefore,shownomultilingualityatall.Usingthismetric,wecanconcludethatDMOZisatthefirstposition,followedbyUDC,DDC,LCC,andthenCC.

Consistencymeasuresthelevelofuniformityinclassificationsystemtoclassifysubjects.Accordingto(Batty,1967),intheearlierstages,CCshowsnoconsistencybutbytheadditionofconsistencycannons,ithasgraduallybecomeconsistent.LCCseemslessconsistentinexpressingdifferentsubjectsareas(Madge,2011).DDCandLCCwerefoundshortofdefiningandclassifyingreligiousholdingsespeciallyJewishcontents.Theseschemesalsoshowbiasnesstowardsdifferentreligiousandregionalcontents(Maddaford&Briefing).AlthoughDDCisalittlebitinconsistent,stillitcanclassifycomplexsubjects(Gnolietal.,2016).UDCalsoshowsinconsistency,whichcanbesortedoutbyintroducingspecificUDCclassestodatabaseinonlinesystem(Kaosar,2008).DDCshowscomparativelygreaterconsistencyinclassifyingnewsubjectswithconstantuniformity;CCisrankedsecondbecauseoftheintroductionofcanonsofconsistency.LCCandUDCarerankedatthirdposition.Forbeingonlylimitedtothescientificresearcharticles,IEEE


60

andACMareatfourthposition.DMOZstandsatfifthpositionduetoitslooseboundariesofcategorization.

Theinteroperabilitydetermineshowmuchagivenclassificationschemeisinteroperableinexpressingitsclassificationartifactswithotherschemes.UDCisinteroperable(Koch&Day,1997)andsupportsintegrationwithothersystems.CC,becauseofitssub-continentalorigin,showslimitedinteroperability(ANeelameghan&Lalitha,2013)(Raghavan,2016).LCCshowsinteroperabilitybybeingcapabletomapwithDDC(Vizine-Goetz,2009).TheinteroperabilityandmultilingualityofDDCenablesittomapwithotherclassificationschemes(Vizine-Goetz,2009).IEEE,ACMandDMOZdatasetsareinteroperablewithotherwebapplications.Basedonthismetric,DDC,LCC,UDC,ACM,IEEEandDMOZarestandingatfirstpositionbecauseofthepresenceoftheirinteroperabilityanddataharvestingprotocolsandontologiesinthedigitalenvironment.DMOZstandsatthesecondpositionbecauseoflimitedinteroperability.CCprovidesonlyphilosophicalandtheoreticalmodelbutwefoundnopracticalweb-basedapplicationsoitisnotincludedinthislist.

Byenablingsemanticsearch,aclassificationschemecanproactivelyrespondtoinformationseekersusingitsfacetedstructure.UDC,becauseofitssemanticstructure(Slavic,2008),containssemanticsearchcapability.TheclassificationtheoryandphilosophyofCCprovidesthebasisforclassificationontologydevelopment(Panigrahi&Prasad,2005),whichmakesobviousitscapabilityofsemanticsearchandinference.LCCsupportssemanticsearchthroughLODsupport,semanticallyenabledLCSHandauthoritycontrolfiles("LCLinkedDataService:AuthoritiesandVocabularies,")(Harper&Tillett,2007).DDCalsocontainssemanticfeatures(Green,2015),whichcanbeutilizedinthesemanticsearchapplications.Therefore,itcanbeconcludedthatsemanticsearchisalsosupportedbyDDC.Thismetriccanbebetteranalyzedinthedigitalenvironmentandespeciallythroughanalyzingthesebibliographicclassificationsfortheirontologies.LCCcouldberankedfirstbecauseofitsexpressiveontologywithefficientsemanticsearchapplication.DDCisatsecondposition,becauseofefficientsearchbutlimitedusageofitsontology.ACMisatthirdpositionbecauseofitsexpressiveontologyandefficientsearchbutlimitedcoveragetoscientificdomain.IEEEisatfourthpositionbecauseofitsfacetedsemanticsearch.UDCcomesatfifthpositionbecauseofitsontologicalpresencebutwithlimitedusage.CChasnoapplicationinthedigitalenvironment,whichcoulddemonstrateitscapabilityforsemanticsearch,althoughitprovidesthebasisforthesemanticlevelforallbibliographicclassificationsystems.DMOZlacksinsemanticsearch,whereitisonlybasedonkeywords.

Biasinsubjectrepresentationmeansinclinationfororagainstsomesubjectswhichresultsinunfair,partialnegligenceorfullyignoringanysubject.DDCandLCCarebiasedinrepresentingdifferentknowledgeandregionalinformation,e.g.,Anglo-Americanbias(Tomren,2003),whileUDCisbiasedtowardsEuropeanculture(Fandino,2008).CCisbiasedtowardsdifferentknowledgeareas(Satija&Singh,2010).Aclassificationsystemwithleastbiasnessisrankedhigher.Therefore,inthisconnection,DMOZisrankedhigherforshowingno/leastbiasness;CCisrankedsecondbecauseofthepresenceofacutebiasnessfollowedbyDDCshowingcomparatively


61

lessbiasnesstowardsreligionandregionalsubjects.LCCcomesatthefourthpositionfollowedbyIEEEandACMthatshowgreaterbiasnesstowardscertaindomains.

Enumerativestructureexhibitstherigidhierarchies.LCCisenumerative(Gohetal.,2009;Perles,1995)(Bryant,October4,1993).UDCisnearlyenumerativeandfaceted(Kaosar,2008)(Bryant,October4,1993)andDDCisbothanalytico-syntheticandenumerative(Hallows,2014).CCisfaceted(Chatterjee,2016;Dawson,Brown,&Broughton,2006).Bycomparingthesesystems,LCCfullysupportsenumerativestructure,andthencomesDDC,whereasUDCisnearlyenumerativeandCCshowsnoenumerativestructureatall.LCCandDDCareenumerative.Thetrendistowardssemanticandfacetedstructure,andtherefore,enumerativestructureinclassificationsystemsisnotadesirablecharacteristic.Therefore,thesystemwithenumerativenaturewillberankedlower.Basedonthismetric,CCandDMOZareleastenumerativeandtherefore,rankedhigher,followedbyIEEEandACMatthesecondposition,thenUDCatthethirdposition,whileDDCandLCCatthelast.

Thefacetedstructuremeansthesemanticallyinterlinkedstructureofcategories,whichcanbemergedandcombinedtogenerateanexpressionforexistingornewconcepts(Svenonius,2000).CCisfaceted(Chatterjee,2016;Dawsonetal.,2006).UDCisanalytico-synthetic(Kaosar,2008)andfollowsthefacetedmethodofCCusingdifferentconnectingsymbolsinmixednotationsandusingsubjectfacetsincludingtimeandspace(Chatterjee,2016).IEEEandACMpossessfacetedstructures.DMOZhasonlyhierarchicalstructureandpredefinedcategories.BasedonthismetricwerankCCfirst,UDCsecond,ACMandIEEEthirdwhileDDCandLCCareenumerativestructures,andtherefore,cannotbeincludedinthelist.

Facetedsearchmeanstonavigateorbrowsethroughthefacetedstructureofafacetedclassificationscheme.Facetedsearchisalsoappliedbyselectingdifferentrangesandchoicesfromdifferentfacetsthataregivenbyanyfacetedsystemtosearchtherequiredcontents.ItisdifferentfromsearchcomplexityinthesensethatitlooksatthepatternandcriteriaofsearchthatexistinanyclassificationschemeeitherinthereOPACsorwebapplications.ThetheoryandphilosophyofCCsupportsfacetedsearch&browsingeconomically(Kong,2016),however,tothebestofourknowledge,noreal-worldapplicationdemonstratesitsusefulness.UDCisbasedonthefacetedapproach,whichsupportsfacetedsearch(Tunkelang,2009).LCCsupportsfacetedsearchwiththehelpofLCSH(McGrath,2007).LCCalsoprovidesfacetedsearchthroughtheFacetedApplicationofSubjectTerminology(FAST)application("FacetedApplicationofSubjectTerminology,"2017).DDCprovidesthefacetedsearchthroughtheOCLCClassify16application.Usingthismetric,theseclassificationschemescanberankedasDDCatfirstpositionbecauseDDCisadoptingthefacetedapproachalongwithitsnativeenumerativenatureandstate-of-the-artwebbasedsearchapplicationsdevelopedbyOCLC.LCCisatsecondpositionbecauseofitswebbasedsearchapplicationsanditsadaptationofcomparativelyrestrictedfacetedapproach.IEEE,forprovidingextensivechoiceofsearchingpatterns,standsatthethirdposition.ACMhaspoly-hierarchicaland

16http://classify.oclc.org/classify2/


62

multi-facetedclassificationstructurealongwithrobustsearchmechanism;therefore,itisonfourthpositioninthislist.ThereareverylimitedfacetedsearchapplicationsofUDCandthereforeitstandsatfifthposition.DMOZhashierarchicalstructureinwhichtherequiredelementcanbeaccessedthroughakeywordsearch.Therefore,itisnotprovidinganyfacetedsearch.CChasnosearchapplicationsthatcouldconfirmitssupportforthefacetedsearch.

LODdatasetsmeanstheavailabilityofdatasetsofagivenclassificationsystemonLODcloud.Amongourchoiceofwell-knownclassificationsystemsUDC,LCC,DDC,IEEE,ACMandDMOZhavedatasetsintheLODcloudwhereasCChasnosuchdatasets.ThedefinitionsofclassesandpropertiesaregatheredinLinkedDataVocabularies(LOV),whichareusedfordescribingdifferenttypesofobjectsusedinLODcloud.Thesedefinitionsofdifferentthingsprovidevocabulariesforlinkingthelinkeddata(Foundation,2017).CC,UDC,DDC,LCC,IEEEandDMOZhavenoLOV,whereasACMhasLOVvocabularies.

Themetric“platforms”intheevaluationframework,considerstheapplicabilityofagivenclassificationsysteminreal-worldwebapplicationsandotherdigitalenvironments.Inthisregard,UDCissupportedbyUDCconsortium,DDCbyOCLC,LCCbyLibraryofCongress,ACMbyACMdigitallibrary,IEEEbyIEEEXploredigitallibrary,andDMOZbyOpenDirectoryProject.Tothebestofourknowledge,CChasnotbeenusedbyanyoftheonlineapplications.

Table3.RankingandAverageRankingofClassificationSchemes

Thewarrantsofclassificationworkasauthoritativeactsforclassificationiststoperformthecognitivepracticefordesigningtheclassesandconceptsintheclassificationsystem,theirstructuralpropertiesandthenputtingsubjectsinthespecifiedclasses(Beghtol,1986).CCand

Ranking

StructuralCom

plexity

Notationalbrevity

PredefinedStructure

RulesCom

plexity

TheoreticalLaw

s

Mnemonics

Hospitality

SearchCom

plexity

Usability

Precisionand

Accuracy

Multilinguility

Interoperability

SemanticSearch

Biasness

Enum

erativeStructure

FacetedStructure

FacetedSearch

Consistency

LODdatasets

LOVSupport

AverageRanking

CC 1 2 7 1 5 2 6 2 2 4 2 1 7 5 4 4 1 4 1 1 3.1

UDC 2 3 4 4 3 6 5 3 3 3 5 3 3 3 2 3 2 5 2 1 3.25

DDC 4 5 3 3 4 7 4 4 5 2 4 3 6 4 1 1 6 3 2 1 3.6

LCC 3 4 2 2 2 5 3 4 4 2 3 3 2 1 1 1 5 3 2 1 2.65

ACM 6 7 6 5 1 4 2 4 5 2 1 3 5 2 3 2 3 2 2 1 3.3

IEEE 5 6 5 5 1 3 2 4 5 2 1 3 4 2 3 2 4 2 2 2 3.15

DMOZ 7 1 1 5 1 1 1 1 1 1 6 2 1 6 4 1 1 1 2 1 2.25


63

UDCuseliterarywarrant;DDCandLCCuseliterary&scientificwarrants.ACMandIEEEusescientificresearchwarrant,whileDMOZexhibitsnowarrantofclassification.

Intheaboveparagraphs,wecomparedandevaluatedtheselectedclassificationsystemusingtheevaluationmetrics(showninTable2),anddiscussedhowthesesystemscanberankedbasedonagivenevaluationmetric.However,togiveaholisticviewofthiscomparisonandevaluation,weintroducearankingscoreorlevelsrangingfrom1(meaninglowranking,notapplicable,ornotavailable)to7(meaninghighranking)inhowaclassificationschemeisbestamongitscounterpartsinthelist.Itisalsothecasethatforagivenmetric,multiplesystemsmaybelongtothesamerankinglevel.Byassigningtheserankinglevels,Table3comparesthesesystemsbasedon20metricsbyexcludingplatformsandwarrantsofclassification.Table3alsoreportstheaveragerankingoftheseclassificationsystems,showingDDCattopwithaveragerankingof3.6,followedbyACM=3.3,andUDC=3.25.ItcanbeconcludedthatDDCandUDCareamongthebestclassificationschemesfordescribingprintedaswellasdigitalcollections,whereasACMisbestforclassifyingdigitalcollectionsbelongingtoComputerSciencedomain.However,ACMclassificationsystemcanbeextendedtoincludeotherdomainsaswell.Figure1illustratesgraphicallythecomparisonandevaluationofthesesystems.

Figure1.ComparisonandRankingofClassificationSystems

Table4presentsthestate-of-the-artbibliographicclassificationontologiesincludingBibliographicontology,LCContology,DDContology,UDContology,andDMOZontology.Someoftheseontologiesweredesignedspecificallyforcertaintargetedapplicationse.g.,ACMontologyforACMdigitallibrary,andLCContologyforLibraryofCongressetc.,whereasothershavemultipleusagescenariosandhavebeenusedbyseveralapplications.Anexampleofsuchgeneral-purpose


64

bibliographicclassificationontologyistheBibliographicOntology17,whichisusedbyseveralbibliographicservicesanddigitallibrariese.g.,digitalobjectidentifier(DOI),Zotero,andLibraryofCongressClassificationNumber(LCCN)permalinkservice(Giasson,2012).Thisevaluationframeworkcomparestheseontologiesbasedontheirsize(intermsofnumberofclasses),usageinthestate-of-the-artapplications,LODsupport,theavailabilityofdatasetsondatahub18,andLOVsupport.BylookingatTable3,ACMshowmorecomprehensivenessintermsofnumberofclasses,triplesandLOVsupport.

ClassificationandCategorizationOntologies

No.ofclasses

Applications LODdatasets

LODdatasetstriples

LOVsupport

Bibliographicontology19 69 LibraryofCongressandBibBase

ü 200000 ü

LCContology20 40+ LibraryofCongress

ü NotGiven û

DDContology21 20+ OCLC ü 402288 û

UDContology22 2,600 UDC23 ü 69,000 û

ACMontology24 1469 ACM ü 12402336 ü

IEEELOMmetadataontology(Casali,Deco,Romano,&Tomé,2013)

9 IEEE25Xploredigitallibrary

ü 91564 ü

DMOZontology26 Notgiven

OpenDirectoryProject

ü Notgiven û

Table4.Comparisonofclassificationandcategorizationontologies

18 https://datahub.io19http://purl.org/ontology/bibo20http://id.loc.gov/21http://dewey.info/22http://udcdata.info/23http://udcdata.info/24http://dl.acm.org/ccs/ccs.cfm25http://ieee.rkbexplorer.com/id/26https://www.dmoz.org/rdf.html


65

Issues&ChallengesinClassificationResearch

Althoughbibliographicclassificationhasbeenpracticedsincetheuseofbooksandtheinceptionoflibrary&informationsciencepractices,furtherresearch&developmenteffortsarerequiredfortomeet theclassificationneedsof thedigitalage.Especially,with thearrivalofdigitalholdings,researchersfaceseveralissuesandchallenges.Forexample,automatictextclassificationperformscategorizationofresourcesusingordinarymetrics includingTF-IDFandclassification in its truesense isyet tobeachieved (Yi,2006).Tohandle the issue, text classification isalsocarriedoutthrough semantic indexing but its accuracy and precision are yet to be achieved. Semantic andstructuralrelationshipsamongdifferentpartsoftextcorpusisstillatinfantlevelandhasnotbeenexploitedtotheirfullestsothatthesecanbeusedintextclassificationinmoremeaningfulways.Other challenges in text classification include handling huge data resulted by applying aclassificationscheme,dynamisminclassification,andstructuredissimilarityamongclassificationschemesalthoughtheyagreeuponsubjectastheprimarycharacteristic.

The biasness in DDC and LCC needs to be resolved. Several revisions and proposals are putforwardforaddressingtheproblemofsystematicknowledgeorganizationandsearchingthroughnaturallanguageterms(Miksa,2007).Therearevariousissuesregardingthestructuralupdates,search&retrievalcriteria,andvisualization(Slavic-Overfield,2005).

TherearetwomainchallengesfortheapplicationofthebibliographicclassificationprinciplesinclassifyingtheWeb.First,theprinciplesofthebibliographicclassificationareformulatedfortheprinteddocuments,whichshouldalsobeapplicabletodigitalcollections.Foraddressingthesechallenges,thereisneedtoapplyandmodifybibliographicclassificationprinciplesindigitalenvironments.Second,itisrequiredtoexploithiddenhierarchiesandconceptstobebetterclassifiedbytheprinciplesofbibliographicclassificationforprecisediscovery,searchandretrieval(J.Mai,2004).

Theissueofdependentprocessofclassificationofanyobjectperpredefinedcriteriaandprinciplesisimportanttoaddressforfindingaplaceinthisageofsearchengines.Thisissuecanbetailoredbytheprinciplesofclassification,sothattheconventionalprinciplesaremodifiedtoconsiderthepurposeofclassificationanddomainofobjects.Forthisissuesemanticwebandontologiescanplayavitalroleinbibliographicclassification,whichcanprovideindependentclassificationofthebibliographicclassificationpredefinedtheories(Hjørland,2012).

Theissueofheterogeneityconflicts,whicharisebecauseoftheinconsistenciesandstructuraldivergences,arethechallengesforthesemanticinteroperability.Semanticinteroperabilitycanbebroughtintothebibliographicrecordsinsidethebibliographicsystemandacrossthesystemsthroughdifferentphasesofinterlinking,evaluation,analysis,remodeling&conversionforanalyzing,andrestructuringthebibliographicdata(Tallerås,2013).

Bibliographicdataisinmulti-format,multi-topical,multi-lingualandmulti-targeted.Fortacklingtheseissues,thebibliographicdatamustbemademutuallyinteroperableformakingit


66

interlinked,searchable,andpresentedinaharmonizedwayacrosstheboundariesofthedatasetsanddatasilos.Theinteroperabilityproblemarisesatthesyntacticlevelformakingconsistentthecharactersets,notations,dataformatsandrecordsindifferentsystems.Theinteroperabilityproblemisalsoarisingatthesemanticlevelbecauseofthedifferenceindatainterpretationsanddifferenceinvocabularies,andprecisionlevelsindataencoding.Publishing,collectingandmaintenanceofbibliographicdatabymultiorganizationsthroughownestablishedstandardsandbestpracticesinWeb2.0(Hyvönen,2012).Withtheseproblemsinhand,thetransitionofthisdatafromsyntacticalWebtoSemanticWebisachallengeforbringingtheuniformityinrecordsthataregeneratedbydiversesources,encodedinmulti-bibliographicsystems,crossbibliographicsystemsinteroperability,thevisualizationofbibliographicdataaccordinglyasperneedfordifferentcontexts.Foraddressingtheseproblemsthereisaneedforcoordinationandcollaborationbetweenbibliographicdatapublishersandthetechnicaldevelopersofthewebapplications(Hyvönen,2012).

Thereisvarietyofmetadatastandardsandschemasfordefining,managing,resourcediscovery,search&retrieval,preserving,mapping,cross-walking,integrity,accuracy,andauthenticityofmetadataandbibliographicdata.Butforthesetaskstobehandledwithgreatsimplicity,semanticrichnessandaccuracy,auniversalallinonemetadataformatandschemaistheneedoftheday(Ramesh,Vivekavardhan,&Bharathi,2015)togetoutofthisjungleofstandards(Gartner,2016).Thisway,themetadatapublishersandmanagerscouldgetrelievedandthejobwillbecomeeconomicintermsoftime,management,andsearch&retrieval.

ThreemaintasksweresetinSemanticPublishingchallenges2015.Thesetasksare:(i)extractingdataonworkshops’qualityindicators;(ii)extractingdataonaffiliations,citations,funding;and(iii)interlinking.Severalchallengeswerefacedwhilefulfillingthesetasks.Thesetasksarebeingfulfilledthroughaproposedsolution,whichiscomposedofatextminingpipeline,LODeXporterandNamedEntityRecognition(NER)fornamedentitiesextractionformtextandlinkingthemtoresourcesontheLODcloud(Sateli&Witte,2015).

In(Peroni,2012)threemainissuesofsemanticpublishingareaddressedwhichare:lackofdocumentpublishinguniversalmetadataschemasaccordingtopublishingvocabulary,lackingofefficientuserinterfacethatarebasedonmodelsandtheoriesofsemanticpublishing,andthereisaneedforatoolthatsemanticallylinkanddescribedocumenttext.Theseissuesrequiretheurgentneedforcomprehensiveontologiesfordocumentpublishingdomain.

(Ferrara & Salini, 2012) tossed 10 challenges for multiple dimensions of data in terms ofbibliographic analysis. These challenges are: (i) analyzing bibliographic data in amultidimensionalpattern;(ii)discoveringandintegratingdatacomingfromdiversesources;(iii)detecting multiple references to the same item and cleaning, normalizing, and disambiguatingbibliographicdatarecords;(iv)analyzingmultidimensionalnatureofbibliographicdatathroughmultivariateanalysis foraggregatingthedata;(v)comparingdifferentelementsofbibliographicdata and its ranking accordingly, (vi) aggregating indexes of different nature with respect to


67

differentparameters,dimensions,andelementsofbibliographicdata;(vii)dealingwithmultipleindexes for thesame itemwith differentvaluescoming fromdifferentsources; (viii)extractingandindexingtextualinformationfromtextcorpusinsupportoftextmining;(ix)analyzingtextualdata topic-wise and describing these topics for research and learning process and tracingdifferent trends; and (x) combining multidimensional information for finding trends inbibliographicdatacollection.

BibliographicclassificationsystemsarebeingincorporatedinLOD.InDewey.info27,aprototypeversionofDDCisdesignedforlinkingitsdatasetinlinkeddatacloud.TheintentionistoprovideaplatformforDDCdataontheWebhavingsummariesoftop3levelsofclassificationorderofDDC22ndeditionin11differentlanguageencodedinRDF/SKOS,havingactionableURIsforeveryclass,representationformachinesisinRDF,andforhumansinXHTML+RDFs,andserializationavailableinformatsofRDF/XML,TurtleandJSON,andwithSPARQLendpoint.(OCLC2011;MitchellandPanzer2013).However,thisversionofDDConLODcloudisstillatinfantstagetocoverdifferentsubjectsandtobewidelyusedingeneratingandcreatingdocumentsmetadata.

LibraryofCongressLinkedDataserviceprovidesaccesstocommonlyusedstandardsandvocabulariesdevelopedbyLibraryofCongress.Thisincludesdatavalues,controlledvocabularies,andpreservationvocabularieswhicharepartofthisservice.ThisserviceprovidesaccesstoLCSH,LCCnameauthorityfiles,LCC28,LCchildren'ssubjectheadings,LCgenre/formterms,thesaurusforgraphicmaterials,MARCrelators,MARCcountries,MARCgeographicareas,MARClanguages,ISO639-1languages,ISO639-2languages,ISO639-5languages,extendeddate/timeformat,preservationevents,andpreservationlevelroleandcryptographichashfunctions.TheauthoritiesandvocabulariescurrentlyincludedinthisservicearelistedontheLinkedDataservice(LibraryofCongress2014).However,itlacksinvocabulariesforsupportingPREMIS,MARC,MODS,METS,andMIX.

As presented in Section 2, several ontologies have been developed for describing and sharingknowledge about bibliographic classification. However, the available ontologies are limited inseveralwayse.g.,theseontologiesarenotthecompleteclonesofclassificationschemesofwhichthey are deemed to be ontologies and they also not mature enough in terms of metadatacollection. In addition, these ontologies still couldn’t break the cross-classification schememetadatacollectionbarriersi.e.,theyarenotinteroperableenoughtoharvestthemetadataacrossbibliographic ontology system. Therefore, further initiatives are required to develop maturedbibliographicontologieswhichfullyclonebibliographicschemesthatareinpracticaluseandhavestrong theoretical ground. These ontologies must be interoperable and sharing metadatacollectionwithotherbibliographicontologies. Inthisway in futurewecanhaveontology-basedgeneral bibliographic classification system by fusion of the new and existing bibliographicontologiesforbettermanagementoftheknowledgeartifacts. 27https://datahub.io/dataset/dewey_decimal_classification


68

CONCLUSIONS

Withthearrivalofdigitalcollections,newchallengesofpreservation,curationaswellasresourcediscovery&access(retrieval)haveemergedthatneedsproperattention,whereclassificationschemesandontologiescanplayasignificantrole.BycomparingandevaluatingtheavailablebibliographicclassificationandcategorizationsystemsitisconcludedthatcurrentlyDDCisthebestclassificationsystemfollowedbyUDC,andACM.Thebibliographicclassificationontologiesarelimitedinonewayortheothere.g.,someofthesearecomprehensivelikeUDCandACMbutlacksupportforLODandLOVetc.,whileotherssupporttheselateraspectsbutlackcomprehensiveness.Keepinginviewtheavailablebibliographicclassificationontologiesandtheirlimitations,werecommendthatauniversalbibliographicclassificationontologyshouldbedevelopedbyusingtheclassesfromtheavailableontologiesandprovidingsupportintermsofavailabilityofdatasets,supportforinteroperability,LOD,andLinkeddatavocabularies.

Fordevelopingamoremeaningfulclassificationsystem,equallyapplicabletodigitalenvironments,itisnecessarytoconsiderthebookstructuralsemanticssuchastableofcontents,headings,chapters,sections,subsections,figures,algorithms,mathematicalequations,quotationsetc.,andthelogicalconnectionsincontents(Khusro&Ullah,2016;I.Ullah&Khusro,2016)aswellasaboutthebookinformationi.e.,thebibliographicdetailsoftheholdings.Tomeet,theformerrequirement,acomprehensiveontologylikeBookOnt(A.Ullah,Ullah,Khusro,&Ali,2016)couldbeused,whichcanbemappedwithanybibliographicontologylikee.g.,BibliographicOntology29.However,astheevaluationframeworkssuggest,DDC,UDC,andACMClassificationSystemshouldbeexploitedindesigningsuchageneral-purposeclassificationsystem.

REFERENCES

The2012ACMComputingClassificationSystem.RetrievedMarch20,2017,fromhttp://www.acm.org/about/class/2012

AboutUniversalDecimalClassification(UDC).RetrievedMarch21,2017,fromhttp://www.udcc.org/index.php/site/page?view=about

Albrechtsen,H.(2000).Whowantsyesterday'sclassifications?Informationscienceperspectivesonclassificationschemesincommoninformationspaces.InK.Schmidt(Ed.),Papers.TechnicalUniversityofDenmark,CenterforTele-Information.

Batley,S.(2014).ClassificationinTheoryandPractice.Oxford:ChandosPublishing.

Batty,C.D.(1967).Anintroductiontocolonclassification:ArchonBooks.

Beghtol,C.(1986).Semanticvalidity:conceptsofwarrantinbibliographicclassificationsystems.Libraryresources&technicalservices,30(2),109-125.

29http://bibliontology.com/#


69

Beghtol,C.(1998).Knowledgedomains:multidisciplinarityandbibliographicclassificationsystems.KnowledgeOrganization,25(1-2),1-12.

Beghtol,C.(2001).RelationshipsinclassificatorystructureandmeaningRelationshipsintheorganizationofknowledge(pp.99-113):Springer.

Berners-Lee,T.(2006,June18,2009).Linkeddata.DesignIssues.RetrievedMarch21,2017,fromhttps://www.w3.org/DesignIssues/LinkedData.html

Betts,T.,Milosavljevic,M.,&Oberlander,J.(2007).TheutilityofinformationextractionintheclassificationofbooksAdvancesinInformationRetrieval(pp.295-306):Springer.

Bizer,C.,Heath,T.,&Berners-Lee,T.(2009).Linkeddata-thestorysofar.SemanticServices,InteroperabilityandWebApplications:EmergingConcepts,205-227.

Boykin,J.(2016).AssessingDMOZ:AQualityReview.Retrieved14-03-2016,2016,fromhttps://www.seochat.com/c/a/search-engine-news/assessing-dmoz-a-quality-review/

Bryant,B.(October4,1993).'NumbersYouCanCountOn'DeweyDecimalClassificationIsMaintainedatLC.LibraryofCongressInformationBulletin,52(18).http://www.loc.gov/loc/lcib/93/9318/count.html

Buchanan,B.(1979).Theoryoflibraryclassification.

Campbell,D.G.(2002).CentripetalandCentrifugalForcesinBibliographicClassificationResearch.PaperpresentedattheASISSIG/CRClassificationResearchWorkshop.

Casali,A.,Deco,C.,Romano,A.,&Tomé,G.(2013).Anassistantforloadinglearningobjectmetadata:Anontologybasedapproach.

Chan,L.M.(2000).ExploitingLCSH,LCC,andDDCtoRetrieveNetworkedResources:IssuesandChallenges.

Chan,L.M.,Intner,S.S.,&Weihs,J.(2016).GuidetotheLibraryofCongressclassification:ABC-CLIO.

Chapman,J.W.,Reynolds,D.,&Shreeves,S.A.(2009).Repositorymetadata:approachesandchallenges.Cataloging&classificationquarterly,47(3-4),309-325.

Chatterjee,A.(2016).UniversalDecimalClassificationandColonClassification:Theirmutualimpact.AnnalsofLibraryandInformationStudies(ALIS),62(4),226-230.

Cliff,P.(2008).JISC-Repositories:SubjectClassificationThreadSummary.

Comaromi,J.P.,&Satija,M.P.(1983).BrevityofnotationinDeweydecimalclassification:Metropolitan.

Dawson,A.,Brown,D.,&Broughton,V.(2006).Theneedforafacetedclassificationasthebasisofallmethodsofinformationretrieval.PaperpresentedattheAslibproceedings.


70

DeGrolier,E.(1962).Astudyofgeneralcategoriesapplicabletoclassificationandcodingindocumentation.

DeweyDecimalClassificationsummaries.RetrievedMarch21,2017,fromhttps://www.oclc.org/en/dewey/features/summaries.html

DeweyServices:DeweyDecimalClassificationSystem.RetrievedMarch20,2017,fromhttps://www.oclc.org/content/dam/oclc/services/brochures/211422usb_dewey_services.pdf

Dorji,T.C.,Atlam,E.-s.,Yata,S.,Fuketa,M.,Morita,K.,&Aoe,J.-i.(2011).Extraction,selectionandrankingofFieldAssociation(FA)Termsfromdomain-specificcorporaforbuildingacomprehensiveFAtermsdictionary.KnowledgeandInformationSystems,27(1),141-161.doi:10.1007/s10115-010-0296-x

Dousa,T.M.(2009).EvolutionaryorderintheclassificationtheoriesofCACutter&ECRichardson:itsnatureandlimits.

Encyclopedia,N.W.(1August2014).Libraryclassification.2017,fromhttp://www.newworldencyclopedia.org/entry/Library_classification

FacetedApplicationofSubjectTerminology.(2017).RetrievedMarch21,2017,fromhttp://www.oclc.org/research/themes/data-science/fast.html

Fandino,M.(2008).UDCorDDC:anoteaboutthesuitablechoicefortheNationalLibraryofLiechtenstein.ExtensionsandCorrectionstotheUDC.

Ferrara,A.,&Salini,S.(2012).Tenchallengesinmodelingbibliographicdataforbibliometricanalysis.Scientometrics,93(3),765-785.

Foundation,O.K.(2017).AboutLOV.fromhttp://lov.okfn.org/dataset/lov/about

Francu,V.(2007).Multilingualaccesstoinformationusinganintermediatelanguage:ProefschriftvoorgelegdtothetbehalenvandegraadvandoctorindeTaal-enLetterkundeaandeUniversiteitAntwerpen.

Gartner,R.(2016).Metadata:Springer.

Giasson,B.D.A.F.(2012).ProjectsusingBIBO.fromhttp://www.bibliontology.com/projects.html

Giess,M.D.,Wild,P.,&McMahon,C.(2007).Theuseoffacetedclassificationintheorganisationofengineeringdesigndocuments.PaperpresentedattheProceedingsoftheInternationalConferenceonEngineeringDesign2007.

Gilchrist,A.(2015).ReflectionsonKnowledge,CommunicationandKnowledgeOrganization.KnowledgeOrganization,42(6),456-469.

Giunchiglia,F.,Marchese,M.,&Zaihrayeu,I.(2007).EncodingclassificationsintolightweightontologiesJournalondatasemanticsVIII(pp.57-81):Springer.


71

Gnoli,C.,Merli,G.,Pavan,G.,Bernuzzi,E.,&Priano,M.(2008).FreelyfacetedclassificationforaWeb-basedbibliographicarchive:theBioAcousticReferenceDatabase.

Gnoli,C.,Pusterla,L.,Bendiscioli,A.,&Recinella,C.(2016).Classificationforcollectionsmappingandqueryexpansion.

Goh,Y.M.,Giess,M.,McMahon,C.,&Liu,Y.(2009).Fromfacetedclassificationtoknowledgediscoveryofsemi-structuredtextrecordsFoundationsofComputational,IntelligenceVolume6(pp.151-169):Springer.

Green,R.(2015,October29-30,2015).Relationalaspectsofsubjectauthoritycontrol:thecontributionsofclassificatorystructure.PaperpresentedattheProceedingsoftheInternationalUDCSeminar2015Classification&authoritycontrolExpandingresourcediscovery,Lisbon.

Hallows,K.M.(2014).It'sAllEnumerative:ReconsideringLibraryofCongressClassificationinUSLawLibraries.LawLibr.J.,106,85.

Harper,C.A.,&Tillett,B.B.(2007).LibraryofCongresscontrolledvocabulariesandtheirapplicationtotheSemanticWeb.Cataloging&classificationquarterly,43(3-4),47-68.

Hjorland,B.(1999).TheDDC,theuniverseofknowledge,andthepost-modernlibrary.JournaloftheAssociationforInformationScienceandTechnology,50(5),475.

Hjørland,B.(2007).Semanticsandknowledgeorganization.Annualreviewofinformationscienceandtechnology,41(1),367-405.

Hjørland,B.(2008).Coreclassificationtheory:areplytoSzostak.JournalofDocumentation,64(3),333-342.

Hjørland,B.(2012).IsclassificationnecessaryafterGoogle?JournalofDocumentation,68(3),299-317.

Hjørland,B.(2013).Theoriesofknowledgeorganization—theoriesofknowledge:KeynoteMarch19,2013.13thMeetingoftheGermanISKOinPotsdam.KnowledgeOrganization,40(3),169-181.

Hjørland,B.(2016).Subject(ofdocuments).KnowledgeOrganization,44(1),55-64.

Hyman,R.J.(1980).Shelfclassificationresearch:past,present--future?Occasionalpapers(UniversityofIllinoisatUrbana-Champaign.GraduateSchoolofLibraryScience);no.146(Nov.1980).

Hyvönen,E.(2012).Publishingandusingculturalheritagelinkeddataonthesemanticweb.SynthesisLecturesontheSemanticWeb:TheoryandTechnology,2(1),1-159.

Jacob,E.K.(2004).Classificationandcategorization:adifferencethatmakesadifference.Librarytrends,52(3),515.


72

Jonassen,D.H.(2004).Handbookofresearchoneducationalcommunicationsandtechnology:Taylor&Francis.

Jones,K.S.(1970).Somethoughtsonclassificationforretrieval.JournalofDocumentation,26(2),89-101.

Joorabchi,A.,&Mahdi,A.E.(2009).Leveragingthelegacyofconventionallibrariesfororganizingdigitallibraries.PaperpresentedattheInternationalConferenceonTheoryandPracticeofDigitalLibraries.

Joorabchi,A.,&Mahdi,A.E.(2011).Anunsupervisedapproachtoautomaticclassificationofscientificliteratureutilizingbibliographicmetadata.J.Inf.Sci.,37(5),499-514.doi:10.1177/0165551511417785

Kaosar,A.(2008).Merit&DemeritofusingUniversalDecimalClassificationontheInternet.

Kaula,P.(1965).ColonClassification:GenesisandDevelopment.LibraryScienceToday.Ranganathan’sFestschrift,1,87-93.

Khusro,S.,&Ullah,I.(2016).Towardsasemanticbooksearchengine.Paperpresentedatthe2016InternationalConferenceonOpenSourceSystems&Technologies(ICOSST'16),Lahore,Pakistan.

Koch,T.,&Day,M.(1997).DESIRE-DevelopmentofaEuropeanServiceforInformationonResearchandEducation.

Koch,T.,Day,M.,Brümmer,A.,Hiom,D.,Peereboom,M.,Poulter,A.,&Worsfold,E.(1997).TheroleofclassificationschemesinInternetresourcedescriptionanddiscovery.WorkPackage,3.

Kong,W.(2016).ExtendingFacetedSearchtotheOpen-DomainWeb.UniversityofMassachusettsAmherst.

Koshman,S.(1993).Categorizationandclassificationrevisited:areviewofconceptinlibraryscienceandcognitivepsychology.CurrentStudiesinLibrarianshipSpring/Fall,26.

Kwaśnik,B.H.,&Rubin,V.L.(2003).Stretchingconceptualstructuresinclassificationsacrosslanguagesandcultures.Cataloging&classificationquarterly,37(1-2),33-47.

Kyle,B.,&Vickery,B.C.(1961).TheUniversalDecimalClassification:presentpositionandfuturedevelopments:Unesco.

LCLinkedDataService:AuthoritiesandVocabularies.Retrieved28Feb2017,2017,fromhttp://id.loc.gov

Lee,H.-L.(2012).EpistemicfoundationofbibliographicclassificationinearlyChina:ARuclassicistperspective.JournalofDocumentation,68(3),378-401.

LibraryofCongressClassification.(2014,10/1/2014).RetrievedMarch20,2017,fromhttps://www.loc.gov/catdir/cpso/lcc.html


73

LibraryofCongressClassificationOutline:ClassP-LanguageandLiterature.[Pressrelease].Retrievedfromhttps://www.loc.gov/aba/cataloging/classification/lcco/lcco_p.pdf

.LibraryofCongressSubjectHeadings:Pre-vs.Post-CoordinationandRelatedIssues.(March15,2007)ReportforBeacherWiggins,Director,Acquisitions&BibliographicAccessDirectorate,LibraryServices,LibraryofCongress(pp.49).CatalogingPolicyandSupportOffice.

LibraryOPACscontainingUDCcodes.RetrievedMarch21,2017,fromhttp://www.udcc.org/index.php/site/page?view=opacs

Linkeddata.fromhttps://www.w3.org/standards/semanticweb/data

Losee,R.M.(1993).Sevenfundamentalquestionsforthescienceoflibraryclassification.KnowledgeOrganization,20,65-65.

Maddaford,S.,&Briefing,C.LibraryofCongressClassificationSystem.

Madge,O.-L.(2011).EvidenceBasedLibraryandInformationPractice.StudiideBiblioteconomieşiŞtiinţaInformării(15),107-112.

Mai,J.-E.(2003).Thefutureofgeneralclassification.Cataloging&classificationquarterly,37(1-2),3-12.

Mai,J.-E.(2004).Classificationincontext:relativity,reality,andrepresentation.KnowledgeOrganization,31(1),39-48.

Mai,J.-E.(2005).Analysisinindexing:documentanddomaincenteredapproaches.InformationProcessing&Management,41(3),599-611.doi:http://dx.doi.org/10.1016/j.ipm.2003.12.004

Mai,J.-E.(2009).Theboundariesofclassification.

Mai,J.-E.(2010).Classificationinasocialworld:biasandtrust.JournalofDocumentation,66(5),627-642.

Mai,J.-E.(2011).Themodernityofclassification.JournalofDocumentation,67(4),710-730.

Mai,J.(2004).ClassificationoftheWeb:challengesandinquiries.KnowledgeOrganization,31(2),92.

Mancuso,J.(1994).GeneralPurposevsSpecialPurposeCouplings.Paperpresentedatthe23rdTurbomachinerySymposium,Dallas,TX,Sept.

Manning,C.D.,Raghavan,P.,&Schütze,H.(2008).Introductiontoinformationretrieval(Vol.1):CambridgeuniversitypressCambridge.

Maron,M.E.,Kuhns,J.L.,&Ray,L.C.(1959).Probabilisticindexing:astatisticalapproachtothelibraryproblem.PaperpresentedatthePreprintsofpaperspresentedatthe14thnationalmeetingoftheAssociationforComputingMachinery,Cambridge,Massachusetts.


74

McGrath,K.(2007).Facet-basedsearchandnavigationwithLCSH:Problemsandopportunities.code4libJournal,1.

McIlwaine,I.C.(1997).TheUniversalDecimalClassification:Somefactorsconcerningitsorigins,development,andinfluence.JournaloftheAmericanSocietyforInformationScience(1986-1998),48(4),331.

Miksa,S.D.(2007).Thechallengesofchange:areviewofcatalogingandclassificationliterature,2003-2004.Libraryresources&technicalservices,51(1),51.

Neelameghan,A.,&Lalitha,S.(2013).Multilingualthesaurusandinteroperability.DESIDOCJournalofLibrary&InformationTechnology,33(4).

Neelameghan,A.,&Parthasarathy,S.(1997).SRRanganathan'sPostulatesandNormativePrinciples:ApplicationsinSpecializedDatabasesDesign,IndexingandRetrieval:SaradaRanganathanEndowmentforLibraryScience.

Nizamani,S.,Memon,N.,&Wiil,U.K.(2011).ClusterBasedTextClassificationModelCounterterrorismandOpenSourceIntelligence(pp.265-283):Springer.

Painter,A.F.(1974).Classification:TheoryandPractice.DrexelLibraryQuarterly,10(4),n4.

Panigrahi,P.,&Prasad,A.(2005).InferenceEngineforDevicesofColonClassificationinAI-basedAutomatedClassificationSystem.

Perles,B.(1995).FacetedClassificationsandThesauri.RetrievedfromHowardBesser'sWebwebsite:http://besser.tsoa.nyu.edu/impact/f95/Papers-projects/Papers/perles.html

Peroni,S.(2012).SemanticPublishing:issues,solutionsandnewtrendsinscholarlypublishingwithintheSemanticWebera.alma.

Piros,A.(2014,29February1,2014).AdifferentapproachtoUniversalDecimal

ClassificationinaMechanizedRetrieval

System.PaperpresentedattheProceedingsofthe9thInternationalConferenceonAppliedInformatics

Eger,Hungary.

Pollitt,A.S.(1998).Thekeyroleofclassificationandindexinginview-basedsearching:Technicalreport,UniversityofHuddersfield,UK,1998.http://www.ifla.org/IV/ifla63/63polst.pdf.

Press,O.F.(2002).Introductiontothedeweydecimalclassification.

Raghavan,K.(2016).TheColonClassification:Afewconsiderationsonitsfuture.AnnalsofLibraryandInformationStudies(ALIS),62(4),231-238.

Rahman,A.,&Ranganathan,T.(1962).SeminalMnemonics.AnnalsofLibraryScience,9,53-67.


75

Ramesh,P.,Vivekavardhan,J.,&Bharathi,K.(2015).MetadataDiversity,InteroperabilityandResourceDiscoveryIssuesandChallenges.DESIDOCJournalofLibrary&InformationTechnology,35(3).

Ranganathan,S.R.(1968).Choiceofschemeforclassification.Lib.Sci.withaslanttoDocumentation,5(1),1-69.

Reiner,U.(2008).AutomaticanalysisofDeweydecimalclassificationnotationsDataanalysis,machinelearningandapplications(pp.697-704):Springer.

Rodriguez,R.D.(1984).Hulme'sconceptofliterarywarrant.Cataloging&classificationquarterly,5(1),17-26.

Rosenfeld,L.,&Morville,P.(2002).Informationarchitecturefortheworldwideweb:"O'ReillyMedia,Inc.".

SanSegundoManuel,R.(2008).SomeargumentsagainstthesuitabilityofLibraryofCongressClassificationforSpanishLibraries.ExtensionsandCorrectionstotheUDC.

Sateli,B.,&Witte,R.(2015).AutomaticconstructionofasemanticknowledgebasefromCEURworkshopproceedings.PaperpresentedattheSemanticWebEvaluationChallenge.

Satija,M.P.(2013).ThetheoryandpracticeoftheDeweydecimalclassificationsystem:Elsevier.

Satija,M.P.(2015).Savethenationalheritage:RevisetheColonClassification.

Satija,M.P.,&Martínez-Ávila,D.(2015).Features,FunctionsandComponentsofaLibraryClassificationSystemintheLIStraditionforthee-Environment.JournalofInformationScienceTheoryandPractice,3(4),62-77.

Satija,M.P.,&Singh,J.(2010).ColonClassification(CC)Encyclopediaoflibraryandinformationsciences(Vol.2,pp.1158-1168).

Schallier,W.(2005).SubjectretrievalinOPAC's:astudyofthreeinterfaces.Paperpresentedatthe7thISKO-SpainConference:ThehumandimensionofknowledgeOrganization,Barcelona.

Singapore,N.L.o.(2016).UsabilityontheWeb.fromhttp://www.nlb.gov.sg/resourceguides/usability-on-the-web/

Slavic-Overfield,A.(2005).Classificationmanagementanduseinanetworkedenvironment:thecaseoftheUniversalDecimalClassification.UniversityofLondon.

Slavic,A.(2006).Interfacetoclassification:someobjectivesandoptions.

Slavic,A.(2008).UseoftheUniversalDecimalClassification:Aworld-widesurvey.JournalofDocumentation,64(2),211-228.

Smiraglia,R.P.,&VandenHeuvel,C.(2011).IdeaCollider:Fromatheoryofknowledgeorganizationtoatheoryofknowledgeinteraction.BulletinoftheAmericanSocietyforInformationScienceandTechnology,37(4),43-47.


76

Subjectclassificationschemes.(2015).fromhttp://www.ifla.org/best-practice-for-national-bibliographic-agencies-in-a-digital-age/node/9042

Sukhmaneva,E.(1970).TheProblemsofNotationandFacetedClassification.17(3-4),112-116.

Svenonius,E.(2000).Theintellectualfoundationofinformationorganization:MITpress.

Tallerås,K.(2013).FromManyRecordstoOneGraph:HeterogeneityConflictsintheLinkedDataRestructuringCycle.InformationResearch:AnInternationalElectronicJournal,18(3),n3.

Tennis,J.T.(2008).Epistemology,theory,andmethodologyinknowledgeorganization:towardaclassification,metatheory,andresearchframework.

Tennis,J.T.(2011).Ranganathan'slayersofclassificationtheoryandtheFASDAmodelofclassification.

Thelwall,M.(2009).Synthesislecturesoninformationconcepts,retrieval,andservices.".Introductiontowebometrics:Quantitativewebresearchforthesocialsciences.

Tomren,H.(2003).Classification,bias,andAmericanIndianmaterials.Unpublishedwork,SanJoseStateUniversity,SanJose,California.

Tunkelang,D.(2009).Facetedsearch.Synthesislecturesoninformationconcepts,retrieval,andservices,1(1),1-80.

Ullah,A.,Ullah,I.,Khusro,S.,&Ali,S.(2016,19-21Dec.2016).BookOnt:AComprehensiveBookStructuralOntologyforBookSearchandRetrieval.Paperpresentedatthe2016InternationalConferenceonFrontiersofInformationTechnology(FIT),Islamabad,Pakistan.

Ullah,I.,&Khusro,S.(2016).InSearchofaSemanticBookSearchEngineontheWeb:AreWeThereYet?ArtificialIntelligencePerspectivesinIntelligentSystems(pp.347-357):Springer.

UniversalDecimalClassificationsummary.(2017).fromhttp://www.udcsummary.info/php/index.php?id=67277&lang=en#

Vizine-Goetz,J.S.M.D.(2009).TheDeweyDecimalClassification.EncyclopediaofLibraryandInformationScience.

Wang,J.(2009).AnextensivestudyonautomatedDeweyDecimalClassification.JournaloftheAmericanSocietyforInformationScienceandTechnology,60(11),2269-2286.

Wijewickrema,C.M.,&Gamage,R.(2013).Anontologybasedfullyautomaticdocumentclassificationsystemusinganexistingsemi-automaticsystem.

Xin,R.S.,Hassanzadeh,O.,Fritz,C.,Sohrabi,S.,&Miller,R.J.(2013).PublishingbibliographicdataontheSemanticWebusingBibBase.SemanticWeb,4(1),15-22.

Yelton,A.(2011).ASimpleSchemeforBookClassificationUsingWikipedia.InformationTechnologyandLibraries,30(1),7-15.


77

Yi,K.(2006).Challengesinautomatedclassificationusinglibraryclassificationschemes.PaperpresentedattheProceedingsofworldlibraryandinformationcongress:72ndiflageneralconferenceandcouncil.

Zhu,Z.(2011).ImprovingSearchEnginesviaClassification.UniversityofLondon.