SemanticWebInformationRetrievalSystem
GraduateSchoolofInformationScienceandElectricalEngineeringKyushuUniversity6-1Kasuga-Koen,KasugaFukuoka816-8580,Japan
HaiboYu
FacultyofInformationScienceandElectricalEngineering
KyushuUniversity6-1Kasuga-Koen,KasugaFukuoka816-8580,Japan
TsunenoriMine
FacultyofInformationScienceandElectricalEngineering
KyushuUniversity6-1Kasuga-Koen,KasugaFukuoka816-8580,Japan
MakotoAmamiya
yu@al.is.kyushu-u.ac.jp
ABSTRACT
mine@al.is.kyushu-u.ac.jpamamiya@al.is.kyushu-u.ac.jp
ThesemanticWebandWebservicetechnologieshavepro-videdbothnewpossibilitiesandchallengestoautomaticinformationprocessing.TherearealotofresearchesonapplyingthesenewtechnologiesintocurrentpersonalWebinformationretrievalsystems,butnoresearchaddressesthesemanticissuesfromthewholelifecycleandarchitecturepointofview.Webservicesprovideanewwayforaccess-ingWebresources,butuntilnow,theyhavebeenmanagedseparatelyfromtraditionalWebcontentsresources.Inthisposter,weproposeaconceptualarchitectureforapersonalsemanticWebinformationretrievalsystem.ItincorporatessemanticWeb,Webservicesandmulti-agenttechnologiestoenablenotonlypreciselocationofWebresourcesbutalsotheautomaticorsemi-automaticintegrationofhybridWebcontentsandWebservices.
CategoriesandSubjectDescriptions:H.4.0[Informa-tionSystems]:InformationSystemsApplicationsD.2.11[Soft-wareEngineering]:SoftwareArchitecturesGeneralTerms:Design,Management
Keywords:SemanticWeb,WebServices,WebPortal,In-formationRetrievalSystem
themaddressedthesemanticissuesfromthewholelifecycleofinformationretrievalandarchitecturepointofview.
TheWebservicesmechanismsprovideagoodsolutionforapplicationinteroperabilitybetweenheterogeneousenviron-ments.Thoughitismainlyusedforbusinessprocessesuntilnow,wecanpredicatethatWebserviceswillsoonbeusedbyWebportalsforinformationgathering,displayinganddelivering,andwillplayavitalroleonWebinformationre-trievalactivities.However,thetraditional“Webcontents”andnew“Webservices”havebeenmanagedseparatelyfortheirpublishing,discovering,accessing,andprocessing,andthereisn’tanypersonalWebinformationretrievalsystemmanagedthemalltogetheruntilnow.
2.APPROACH
Inthisposter,weproposeourconceptualarchitectureforapersonalsemanticWebinformationretrievalsystembasedonthefollowingthreemainideas.
First,“allparticipantscontributeforthesemanticde-scriptionconsistently”.Asweknowthatefficientsearch-ingforhighqualityresultisbasedonapertinentmatch-ingbetweenwell-definedresourcesanduserqueries,wherethematchingreflectsuserpreferences,allparticipants(con-sumer,provider,andmediator)needtocontributetothese-manticdescriptionconsistently.Theuserneedstodescribetheirrequirementsandtheproviderneedstodescribetheircapabilitiespreciselyaswell.Themediatorneedstoen-surethatsemanticsarenotlostsightofduringtheprocess-ingandguaranteethesemanticinteroperabilitybyontologymappingifnecessary.Amulti-agentsystemwillfunctionasthemediatorinourarchitecture.
Second,“integratingWebcontentswithWebservices”.InthesemanticWeb,contentsaregivenwell-definedmeaning,bothWebcontentsandWebservicescanbeconsumedbymachines,sotheyhavethecommongroundtobemanagedtogether.Ontheotherhand,Webusersgenerallyonlycareaboutfinalresults,notrealizationdetails,henceaninte-gratedorunifiedmanagementofWebcontentsandWebser-vicesneedtobecarriedoutthroughdifferentlevelsincludingthedescripitonofcapabilitiesandrequirements,querying,discovering,accessingandaggregation.
Third,“providingagatewaytoallinformationthattheuserisinterestedin”.Usersneedtoaccessandprocessavarietyofinternalandexternalinformationbasedontheir
1.INTRODUCTION
Sincetheinformationoverload,Webinformationretrievalsystemsarefacingnewchallengesforhelpingpeoplenotonlylocatingrelevantinformationpreciselybutalsoaccess-ingandaggregatingavarietyofinformationfromdifferentresourcesautomatically.
WithsemanticWebtechnology,Webinformationcanbegivenwell-definedmeaningwhichcanbeunderstoodandprocessedbymachines.Itprovidesnewpossibilitiesforau-tomaticWebinformationprocessing.
Currently,therearealotofresearchessuchasTAP[1]andHaystack[2],whicharetryingtoapplysemanticWebtech-nologiesintopersonalWebinformationretrievalsystems.However,theyonlyaddressedtheproblemsconcerningcer-tainphasesorcertainaspectsofwhattheyinvolved,noneof
Copyrightisheldbytheauthor/owner.
WWW2005,May10–14,2005,Chiba,Japan.ACM1-59593-051-5/05/0005.
974
ingfordifferentlevels.ForthedetailofourWebsitecapa-bilitydescriptionmechanism,onecanrefertodocument[3].
3.2“Myportal”
“Myportal”iscomposedofthreemainfunctionalcompo-nents:corecomponent,consumercomponentandprovidercomponent.Thecorecomponentprovidesbasicsupportforsemantictechnologiesandinformationmanagement.Itcon-sistsof“KnowledgeBase(KB)”,“QueryEngine(QE)”and“InferenceEngine(IE)”.Asaconsumer,itconsistsofauserinterface,userprofile,consumersearchagentcompo-nentsandbringtogetheravarietyofnecessaryinformationfromdifferentresourcesfortheuser.ThesemanticWebcontentsandtheoutputsofWebservicesbasedonthecom-mondomainontologywillbeintegrated,savedandreusedin“Myportal”.Asaprovider,thecontentsandservicesof“Myportal”canbeconsumedbyhumansaswellasma-chines.Thehumancanbetheuserher/himselforotherpermittedpersons,andthemachinecanbelocalorremote.Theinterfacesforbrowsing,searchingandfacilitatingWebcontentsandservicesneedtobeprovided.Forthedetailofour“Myportal”description,onecanrefertodocument[4].
Figure1:AConceptualArchitecture
preferences,notonlytheinformationfromtheWebbutalsotheinformationcreatedandmanagedbythemselves.Soagatewaytoallrelevantinformationisnecessary.Webportalsaretryingtoprovidesuchfunctionsasgateways.However,“noonesizecanfitall”,andpersonalizationforthespecificuserisstronglyrequired.CurrentlymanyWebportalshaveprovidedpersonalizationsuchas“myyahoo”,“myAOL”,buttheirfunctionsarelimitedastheylacksemanticsandseparationfromtheuser’slocalinformation.ThereforeaWebportalwhichisownedandmanagedbythespecificuser,locatedontheuser’slocaldesktoporlocalnetwork,andconstructedbasedonsemanticWebtechnologyisnecessaryforsatisfyingtherequirementsofausertoallhisrelevantinformation.Thatiswhatwecalled“Myportal”[4].Itisjustlikeacompactportalwithmanymorepersonalizationfunctionalities.
3.3Mediator
Themediatorconsistsofaconsumersearchagent(CSA)andprovidersearchagents(PSA).TheCSAreceivesthere-quirementsfromtheuserinterfaceagent,breaksandtrans-formstherequirementsintoformalqueriesbasedonuserpreferences,communicateswithrelevantagents,selectsandinvokesWebservices,integratestheinformationandsendstheresultsbacktotheuserinterfaceagent.ThePSAre-ceivesqueriesfromaCSAandreturnsmatchingresultstotheCSAbasedondifferentpreferencesandrequirements.TheinformationinGIDandWCDsuchasacategorycanbeusedforWebservicesdiscoveryandselection.
Thesemanticinteroperabilitycanberealizedbyusingacommondomainontologyorthroughontologymapping.WethinkitisreasonabletodefineacommondomainontologyforthecommunicationbetweentheCSAandthePSAsandenabletheCSAandPSAstodoontologymappingforacon-sumerandproviderstothecommonontologyrespectivelyifnecessary.
3.ACONCEPTUALARCHITECTURE
OurconceptualarchitectureofpersonalsemanticWebin-formationretrievalsystemcanbeillustratedasshowninFigure1.
EachproviderdescribestheircapabilitiesinwhatwecallaWSCD(WebSiteCapabilityDescription)andisassignedaPSA(ProviderSearchAgent).Eachconsumerdescribesuser’srequirementsincludinghispreferences.Itisassignedaconsumersearchagent(CSA)andalsohasauserinterfaceagent(UIA)forprovidinganintelligentunifiedinterfacetotheuser.TheCSAandPSAwillfunctionasmediatorsbe-tweentheconsumerandtheproviderastheycommunicatewitheachothertofulfillthesearchingandaccessingtask.Theconsumerisconstructedasa“Myportal”providingagatewaytoallrelevantinformation.
4.CONCLUSION
Inthisposter,weaddressedthemainaspectsofase-manticWebinformationretrievalsystemarchitecturetryingtoanswertherequirementsofthenext-generationsemanticWebuser.Inthefuture,wewouldliketofurtherdetailthecomponentsandinterfaces,andimplementaprototypetorevealthepossibilityandeffectivenessofourproposedarchitecture.
3.1Websitecapabilitydescription(WSCD)
WedescribethecapabilitiesofaWebsitebylayers.First,wesemanticallydescribethegeneralcapabilitiesoftheWebsite,andwecallthisa“generalinformationdescription(GID)”.Second,wegiveitsWebcontentscapabilitydescrip-tion(WCD)andWebservicecapabilitydescription(WSD).TherearelinksfromGIDtoWCDandWSDinordertofacilitatethefurthermatchinganduseofWebcontentsandWebservices.Inordertosemanticallydescribethecapa-bilitiesandsupporttheconcreterealizationofservices,weexpresstheservicecapabilitydescriptionintwolayers:“se-manticWebservicedescription”and“concreteWebservicedescription”.Thishierarchicalcapability-describingmecha-nismenablessemanticcapability-describingandmatchmak-
5.REFERENCES
[1]R.Guha,R.McCool,andE.Miller.Semanticsearch.In
ProceedingsofWWW2003,pages700–709,2003.
[2]D.QuanandD.R.Karger.Howtomakeasemanticweb
browser.InProceedingsofWWW2004,pages255–265,2004.[3]H.Yu,T.Mine,andM.Amamiya.Towardsautomatic
discoveryofwebportals-semanticdescriptionofwebportalcapabilities-.InSemanticWebServicesandWebProcessComposition:FirstInternationalWorkshop,SWSWPC2004,July6,2004,LNCS3387/2005,pages124–136.[4]H.Yu,T.Mine,andM.Amamiya.Towardsasemantic
myportal.InThe3rdInternationalSemanticWebConference(ISWC2004)PosterAbstracts,pages95–96,2004.
975
因篇幅问题不能全部显示，请点此查看更多更全内容