Attribute Aware Anonymous Recommender Systems.pdf

Attribute Aware Anonymous Recommender Systems.pdf

ID:49865573

大小:224.05 KB

页数:8页

时间:2020-03-05

Attribute Aware Anonymous Recommender Systems.pdf_第1页
Attribute Aware Anonymous Recommender Systems.pdf_第2页
Attribute Aware Anonymous Recommender Systems.pdf_第3页
Attribute Aware Anonymous Recommender Systems.pdf_第4页
Attribute Aware Anonymous Recommender Systems.pdf_第5页
资源描述:

《Attribute Aware Anonymous Recommender Systems.pdf》由会员上传分享,免费在线阅读,更多相关内容在应用文档-天天文库

1、AttributeAwareAnonymousRecommenderSystemsManuelStritt,KarenH.L.Tso,LarsSchmidt-ThiemeComputer-basedNewMediaGroup,DepartmentofComputerScienceAlbert-Ludwigs-Universit¨atFreiburg,Germany{stritt,tso,lst}@informatik.uni-freiburg.deSummary.Anonymousrecommendersystemsaretheelectronicpe

2、ndanttoven-dors,whoaskthecustomersafewquestionsandsubsequentlyrecommendproductsbasedontheanswers.Inthisarticlewewillproposeattributeawareclassifier-basedapproachesforsuchasystemandcompareittoclassifier-basedapproachesthatonlymakeuseoftheproductIDsandtoanexistingknowledge-basedsyst

3、em.Wewillshowthattheattribute-basedmodelisveryrobustagainstnoiseandprovidesgoodresultsinalearningovertimeexperiment.1IntroductionRecommendersystems(RS)areusedbyonlinecommercialsites,e.g.ama-zon.comandebay.com(giftfinder),tohelpuserstofindproductsthatfittheirpreferences.Theyuseuserp

4、rofilesthatcontainpreferenceindicatorsforspecificproductsortypesofproductstoassesstheinterestacustomermayhaveinproductofferingsandrecommendthoseheisassumedtolikemost.Preferencescaneitherbespecifieddirectlybycustomers,i.e.,byratingprod-ucts,orindirectlyindicatedbytheirbehavior,e.g.,s

5、earchkeywords,productsviewedindetail,productsputinthemarketbasketorproductspurchasedaswellasfrequenciesanddurationsofsuchevents.Ingeneral,therearetwotypesofRS[10]:(i)RSwithuseridentificationthatrequireausertologin(orrelyonotherunsafeuseridentificationmechanismssuchascookies)andthe

6、reforecanaccumulatepreferenceindicators,e.g.,purchasehistoriesovertime.Wheneveracustomerlogsinthesystem,hispastprofileislookedupandusedforrecommending.Newusershavetoprovidesomefront-upinformation,e.g.,someinitialproductratings,beforetheycanusethesystem.Thisissometimescalledthenew

7、userproblem[9].(ii)Anonymousrecommendersystemsthatdonotrequireauseridentifi-cationandthereforedonothaveanyinitialinformationaboutacustomer.Asinthecontextofe-commercedirectratingsandusage-basedpreferenceindica-torsaretootimeconsumingtocollectforsingleuse,othermechanismshaveto2Manu

8、elStritt,KarenH.L.Tso,LarsSchmidt-Thiemebeusedt

当前文档最多预览五页,下载文档查看全文

此文档下载收益归作者所有

当前文档最多预览五页,下载文档查看全文
温馨提示:
1. 部分包含数学公式或PPT动画的文件,查看预览时可能会显示错乱或异常,文件下载后无此问题,请放心下载。
2. 本文档由用户上传,版权归属用户,天天文库负责整理代发布。如果您对本文档版权有争议请及时联系客服。
3. 下载前请仔细阅读文档内容,确认文档内容符合您的需求后进行下载,若出现内容与标题不符可向本站投诉处理。
4. 下载文档时可能由于网络波动等原因无法下载或下载错误,付费完成后未能成功下载的用户请联系客服处理。