您好,欢迎访问三七文档
1、arXiv:nlin/0611054v3[nlin.AO]18Sep2007FrankE.Walter,StefanoBattiston,andFrankShweitzer:AModelofaTrust-basedReommendationSystemonaSoialNetworkRevisedManusriptforJAAMAS(September07,2007).AModelofaTrust-basedReommendationSystemonaSoialNetworkFrankE.Walter,StefanoBattiston,andFrankShweitzerChairofSystemsDesign,ETHZurih,Kreuzplatz5,8032Zurih,Switzerlandfewalterethz.h,sbattistonethz.h,fshweitzerethz.hAbstratInthispaper,wepresentamodelofatrust-basedreommendationsystemonasoialnet-work.Theideaofthemo。
2、delisthatagentsusetheirsoialnetworktoreahinformationandtheirtrustrelationshipsto lterit.Weinvestigatehowthedynamisoftrustamongagentsa ettheperformaneofthesystembyomparingittoafrequeny-basedreommendationsystem.Furthermore,weidentifytheimpatofnetworkdensity,prefereneheterogeneityamongagents,andknowledgesparsenesstoberuialfatorsfortheperformaneofthesystem.Thesystemself-organisesinastatewithperformaneneartotheoptimum;theperformaneonthegloballevelisanemergentpropertyofthesystem,ahievedwithoutexpliito。
3、ordinationfromtheloalinterationsofagents.Keywords:ReommenderSystem,Trust,SoialNetwork1IntrodutionandMotivationInreentyears,theInternethasbeomeofgreaterandgreaterimportaneineveryone’slife.Peopleusetheiromputersforommuniationwithothers,tobuyandsellprodutson-line,tosearhforinformation,andtoarryoutmanymoretasks.TheInternethasbeomeasoialnetwork, linkingpeople,organisations,andknowledge [33℄andithastakentheroleofaplatformonwhihpeoplepursueaninreasingamountofativitiesthattheyhaveusuallyonlydoneintherea。
4、l-world.Thisdevelopmentonfrontspeoplewithaninformationoverload:theyarefaingtoomuhdatatobeabletoe etively lteroutthepieesofinformationthataremostappropriateforthem.TheexponentialgrowthoftheInternet[21℄impliesthattheamountofinformationaessibletopeoplegrowsatatremendousrate.Historially,peoplehave invarioussituations alreadyhadtoopewithinformationoverloadandtheyhaveintuitivelyappliedanumberofsoialmehanismsthathelpthemdealwithsuhsituations.However,manyofthese,inludingthenotionoftrust,donotyethaveanap。
5、propriatedigitalmapping[23℄.Findingsuitablerepresentationsforsuhoneptsisatopiofon-goingresearh[1;10;17;23;27;29℄.Theproblemofinformationoverloadhasbeeninthefousofreentresearhinomputersieneandanumberofsolutionshavebeensuggested.Theuseofsearhengines[9℄isoneapproah,butsofar,theylakpersonalisationandusuallyreturnthesameresultforeveryone,eventhough1/22FrankE.Walter,StefanoBattiston,andFrankShweitzer:AModelofaTrust-basedReommendationSystemonaSoialNetworkRevisedManusriptforJAAMAS(September07,2007).anyt。
6、wopeoplemayhavevastlydi erentpro lesandthusbeinterestedindi erentaspetsofthesearhresults.Adi erentproposedapproaharereommendationsystems[24;25;26;30℄.Inthefollowing,wepresentamodelofatrust-basedreommendationsystemwhih,inanautomatedanddistributedfashion, ltersinformationforagentsbasedontheagents’soialnetworkandtrustrelationships[15;19;25℄.Trustisatopiwhihhasreentlybeenattratingresearhfrommany elds,inluding,butnotlimitedto,omputersiene,ognitivesienes,soiology,eonomis,andpsyhology.Asaresultofthis,t。
7、hereexistsaplethoraofde nitionsoftrust,somesimilartoeahother,somedi erentfromeahother.Intheontextofourmodel,trustanbede nedastheexpetanyofanagenttobeabletorelyonsomeotheragent’sreommendations.Therearemanyareasofappliationinwhihsuhsystems,orsimilarones,areappliable:someobviousexampleswouldbethefailitiestoshareopinionsand/orratingsthatmanyshoppingorautioningwebsiteso er,butthesamepriniplesofombiningsoialnetworksandtrustrelationshipsanbeappliedinotherdomainsaswell:forexample,inthesienti ommunity,in。
8、formofareommendationsystemforjournal,onferene,andworkshopontributions.ThemodelthatwearegoingtopresentenablesaquantitativestudyoftheproblemandalsoprovidesaskethforasolutionintermsofarealInternetappliation/webservie.Theideaattheoreofthemodelisthatagents•leveragetheirsoialnetworktoreahinformation;and•makeuseoftrustrelationshipsto lterinformation.Inthefollowing,wedesribethemodelandtheresultsobtainedbysimulatingthemodelwithmulti-agentsimulations.Tosomeextent,itisalsopossibletomakeanalytialpreditionso。
9、ftheperformaneofthesystemasafuntionofthepreferenesoftheagentsandthestrutureofthesoialnetwork.Theremainderofthepaperisorganisedasfollows:inthefollowingsetion,weputourworkintotheontextoftherelatedwork.Then,wepresentourmodelofatrust-basedreommen-dationsystemonasoialnetwork.Thisisfollowedbyananalysisoftheresultsfromomputersimulationsandanalytialapproximationsofthemodel.Subsequently,weillustrateanumberofpossibleextensionsandonludewithasummaryofthework.2RelatedWorkReentresearhinomputersienehasdealtwit。
10、hreommendationsystems[26;30℄.Suhsys-temsmostlyfallintotwolasses:ontent-basedmethodssuggestitemsbymathingagentpro leswithharateristisofprodutsandservies,whileollaborative lteringmethodsmeasurethe2/22FrankE.Walter,StefanoBattiston,andFrankShweitzer:AModelofaTrust-basedReommendationSystemonaSoialNetworkRevisedManusriptforJAAMAS(September07,2007).similarityofpreferenesbetweenagentsandreommendwhatsimilaragentshavealreadyhosen[30℄.Often,reommendationsystemsareentralisedand,moreover,theyareo eredbyentit。
本文标题:A Model of a Trust-based Recommendation System on
链接地址:https://www.777doc.com/doc-5065383 .html