合肥工业大学硕士学位论文基于双信息源的协同过滤算法及其应用研究姓名：董全德申请学位级别：硕士专业：计算机应用技术指导教师：王浩2010-10
믹폚쮫탅쾢풴뗄킭춬맽싋쯣램벰웤펦폃퇐뺿 햪 튪 퓚뺺헹늻뛏볓뻧뗄놳뺰쿂ꎬ뗧ퟓ짌컱췸햾늻뛏샻폃룶탔뮯췆볶벼쫵쳡룟탂폃뮧뗄탋좤뫍샏폃뮧뗄훒돏뛈ꆣ킭춬맽싋ꎨCFꎩퟷ캪췆볶쾵춳훐ퟮ돉릦뗄튻쿮벼쫵ꎬ쯼뗄쓜솦틑뺭퓚늻춬뗄뗧ퟓ짌컱쾵춳훐뗃떽쇋퇩횤ꆣ 좻뛸ꎬ뒫춳CF폶떽튻킩벬쫖뗄컊쳢죧샤웴뚯컊쳢ꆢ쫽뻝뗄쾡쫨탔컊쳢ꆢ췆볶뗄뿉뾿탔컊쳢뗈탨튪폐킧뗄뷢뻶냬램ꆣ쳘뇰쫇CF퓚뒦샭붻뮥탔잿ꆢ탨튪쏅벼쓜횪쪶뗄쇬폲ꎬ룼쿔뗃솦늻듓탄ꆣ캪뷢뻶짏쫶컊쳢ꎬ퇐뺿헟틑쳡돶쇋CF폫믹폚쓚죝맽싋뷡뫏뗄믬뫏췆볶벼쫵ꆣ떫쫇ꎬ믹폚쓚죝맽싋탨튪쳡좡헢킩닺욷훖샠뗄탭뛠쳘헷ꎬ헢룶릤ퟷ쫇럇뎣삧쓑뗄ꆣ놾컄헽쫇퓚헢훖놳뺰쿂ꎬ캪쇋뷢뻶킭춬맽싋쯣램훐폶떽뗄쫽뻝쾡쫨탔ꆢ헫뛔뻟폐쏅횪쪶놳뺰뗄쿮쒿췆볶ꎬ쳡돶믹폚쮫탅쾢풴쒣쪽뗄킭춬맽싋쯣램ꆣ ퟷ헟쫗쿈듓뗧ퟓ짌컱뗄췆볶쾵춳죫쫖ꎬ퓄뛁쇋맺쓚췢듳솿쿠맘쇏ꎬ룅쫶쇋뗧ퟓ짌컱훐췆볶쾵춳뗄ퟷ폃뫍펦폃쪵샽ꎻ룸돶뗧ퟓ짌컱췆볶쾵춳뗄쒣탍ꎬ늢뷩짜쇋췆볶쾵춳훐뗄쫤죫쫽뻝뗄훖샠벰웤쳘헷ꎻ뛔뗧ퟓ짌컱훐뗄췆볶쾵춳뗄럖샠뫍펦폃뗄뗤탍벼쫵ퟶ쇋뷏캪짮죫뗄퇐뺿ꆣ 웤뛾ꎬ뛔킭춬맽싋뗄췆볶쯣램뷸탐퇐뺿럖컶ꎬ쮵쏷킭춬맽싋쯣램뗄릤ퟷ풭샭뫍쯣램뗄쫤죫쫤돶ꎻ뛔웕뇩펦폃뗄솽훖킭춬맽싋쯣램ꆪ믹폚폃뮧뗄킭춬맽싋췆볶쯣램뫍믹폚쿮쒿뗄킭춬맽싋췆볶쯣램ꎬ뷸탐쇋뷩짜ꎻ훘뗣쳖싛쇋뒫춳뗄킭춬맽싋쯣램듦퓚뗄컊쳢ꎬ늢럖컶쇋떱잰쳡돶뗄뷢뻶랽램ꎬ횸돶쯼쏇뗄폅쫆뫍늻ퟣ횮뒦ꆣ 웤죽ꎬ쿪쾸뷩짜놾컄쳡돶뗄믹폚쮫탅쾢풴쒣쪽뗄킭춬맽싋쯣램ꎬ룃랽램붫에뛏믮뚯폃뮧뛔쒿뇪쿮쒿뗄탋좤돌뛈붨솢퓚솽룶췆볶ퟩꆪꆪ쿠쯆폃뮧췆볶ퟩꎨퟮ뷼쇚뻓벯뫏ꎩ폫볒췆볶ퟩ믹뒡짏ꎬ냑솽룶췆볶ퟩ뗄붨틩뷡뫏웰살ꎬ탎돉뿉뾿뗄탅쾢풴ꎬ좻뫳ꎬ럖컶룷ퟔ펰쿬믮뚯폃뮧뛔쒿뇪쿮쒿뗄좨훘ꎬ볆쯣믮뚯폃뮧뗄ퟮ훕탋좤뛈ꎬ쪵쿖쾵춳췆볶ꆣ믹폚쮫탅쾢풴쒣쪽뗄킭춬맽싋쯣램ꎨDISCFꎩ돤럖뾼싇뗧ퟓ짌컱훐룶탔뮯럾컱뗄쪵볊쟩뿶ꎬ쪹췆볶쾵춳붨솢퓚룼뿉뾿뗄탅쾢풴믹뒡짏ꎬ럂헦쫔퇩횤쏷룃랽램폐룼뫃뗄췆볶훊솿ꆣ 맘볼듊ꎺ뗧ퟓ짌컱ꎻ룶탔뮯췆볶ꎻ킭춬맽싋벼쫵ꎻ쮫탅쾢풴ꎻ욽뻹뻸뛔욫닮 I
Research on Dual Information Source Model-Based Collaborative Filtering Algorithms Abstract In today’s highly competitive e-commerce environmentꎬ the personalized recommendation has emerged as a critical application which is essential to a Web site to retain visitors and turn casual browsers into potential customersꎮOne of the most successful recommendation techniques is collaborative filteringꎬwhose performance has been proved in various e-commerce applicationsꎮ Howeverꎬconventional CF methods suffer from a few fundamental limitations such as the cold-start problemꎬdata sparsity problemꎬand recommender reliability problemꎮThusꎬthey have trouble dealing with high-involvement knowledge-intensive domainsꎮTo overcome these problemsꎬresearchers have proposed recommendation techniques such as a hybrid approach combining CF with content-based filteringꎮBecause e-commerce Web sites often have various product categoriesꎬextracting the many attributes of these categories for content-based filtering is extremely burdensomeꎮUnder this backgroundꎬ this thesis developed Dual Information Source Model-Based Collaborative Filtering Algorithms (DISCF) to overcome data sparsity problem and in view of knowledge-intensive project recommendationꎮ First of allꎬthe personalized recommendation in e-commerce is discussed and large associated data from domestic and abroad is searchedꎮthis thesis Outlined function of the personalized recommendation system in e-commerce and the application exampleꎬgave the personalized recommendation system's model in e-commerceꎬ introduced types and characteristics of the input data in the recommendation system and has done more thorough research to recommendation system's classification and typical technologyꎮ Secondlyꎬ Collaborative Filtering Algorithms is researchedꎬexplaned principle and I/O of Collaborative Filtering Algorithmsꎻtwo kinds of widely applied Collaborative Filtering AlgorithmsꆪUser-Based and Item-Based recommendation algorithms are introducedꎮThen it is appointed that conventional CF methods suffer from a few fundamental limitationsꎬanalyzed current proposed solutions to these questions and pointed out their superiority and the deficiencyꎮ ThirdlyꎬDual Information Source Model-Based Collaborative Filtering algorithms (DISCF) is discussed detailyꎮthe CF method forms dual recommender II
groups—a similar-users’group and an expert-users’group—as credible information sourcesꎮThenꎬit analyzes each group’s influence on the target customers for the target product categoriesꎮThe DISCF method fully considerated the personalized service's actual situation in e-commerceꎬcaused the recommendation system established in the more reliable information source foundationꎬthe simulation testing proved that this method has the better recommendation qualityꎮ Key wordsꎺElectronic Commerce ꎻPersonalized RecommendationꎻCollaborative FilteringꎻDual Information Source ꎻMAE(mean absolute error) III
닥 춼 쟥 떥 춼 뗧ퟓ짌컱훐췆볶쾵춳쒣탍...............................................................7 춼 킭춬맽싋릤ퟷ풭샭........................................................................12 춼 DISCF쯣램쇷돌춼.........................................................................21 춼 춼쫩뗄쒣쓢쫽뻝벯........................................................................26 춼 뇊볇놾뗄쒣쓢쫽뻝벯....................................................................27 춼늻춬쫽뻝쾡쫨뛈붻뮥탔잿뗄뇊볇놾췆볶킧맻........................................30 춼늻춬쫽뻝쾡쫨뛈붻뮥탔잿뗄춼쫩췆볶킧맻............................................30 VII
뇭 룱 쟥 떥 뇭 룷훖췆볶벼쫵...............................................................................2 뇭 췆볶쾵춳펦폃뻙샽........................................................................6 뇭 뗧ퟓ짌컱훐췆볶쾵춳쫤죫쫽뻝쒣탍..............................................8 뇭 믬뫏췆볶쾵춳벼쫵......................................................................11 뇭 폃뮧-쿮쒿움럖뻘헳.....................................................................13 뇭 A taxonomy of collaborative filtering approach............................17 뇭 User(폃뮧뇭)뇭뷡릹....................................................................25 뇭 Item(쿮쒿뇭)뇭뷡릹....................................................................25 뇭 Rate(폃뮧움럖뇭뷡릹)................................................................25 뇭 럖샠ힼ좷뛈쪾틢뇭......................................................................28 뇭 DISCF쾵춳폫죽훖뗤탍뗄킭춬맽싋쯣램춳볆ힼ좷탔뇈뷏..........28 뇭 DISCF쾵춳폫죽훖뗤탍뗄킭춬맽싋쯣램췆볶ힼ좷탔뇈뷏..........29 VIII
뛀 뒴 탔 짹 쏷 놾죋짹쏷쯹돊붻뗄톧캻싛컄쫇놾죋퓚떼쪦횸떼쿂뷸탐뗄퇐뺿릤ퟷ벰좡뗃뗄퇐뺿돉맻ꆣ뻝컒쯹횪ꎬ돽쇋컄훐쳘뇰볓틔뇪힢뫍훂킻뗄뗘랽췢ꎬ싛컄훐늻냼몬웤쯻죋틑뺭랢뇭믲킴맽뗄퇐뺿돉맻ꎬ튲늻냼몬캪믱뗃 뫏럊릤튵듳톧 믲웤쯻뷌폽믺릹뗄톧캻믲횤쫩뛸쪹폃맽뗄닄쇏ꆣ폫컒튻춬릤ퟷ뗄춬횾뛔놾퇐뺿쯹ퟶ뗄죎뫎릱쿗뻹틑퓚싛컄훐ퟷ쇋쏷좷뗄쮵쏷늢뇭쪾킻틢ꆣ 톧캻싛컄ퟷ헟잩쏻ꎺ뚭좫뗂 잩ퟖ죕웚ꎺ 2010쓪 11퓂 20 죕 톧캻싛컄냦좨쪹폃쫚좨쫩 놾톧캻싛컄ퟷ헟췪좫쇋뷢 뫏럊릤튵듳톧 폐맘놣쇴ꆢ쪹폃톧캻싛컄뗄맦뚨ꎬ폐좨놣쇴늢쿲맺볒폐맘늿쏅믲믺릹쯍붻싛컄뗄뢴펡볾뫍듅엌ꎬ퓊탭싛컄놻닩퓄뫍뷨퓄ꆣ놾죋쫚좨 뫏럊릤튵듳톧 뿉틔붫톧캻싛컄뗄좫늿믲늿럖쓚죝뇠죫폐맘쫽뻝뿢뷸탐볬쯷ꎬ뿉틔닉폃펰펡ꆢ쯵펡믲즨쏨뗈뢴훆쫖뛎놣듦ꆢ믣뇠톧캻싛컄ꆣ (놣쏜뗄톧캻싛컄퓚뷢쏜뫳쫊폃놾쫚좨쫩) 톧캻싛컄ퟷ헟잩쏻ꎺ 뚭좫뗂 떼쪦잩쏻ꎺ췵뫆 잩ퟖ죕웚ꎺ 2010쓪11 퓂21 죕 잩ퟖ죕웚ꎺ2010 쓪 11퓂25죕 톧캻싛컄ퟷ헟뇏튵뫳좥쿲ꎺ 릤ퟷ떥캻ꎺ쯞훝톧풺탅쾢릤돌톧풺 뗧뮰ꎺꎨ0557ꎩ3680439 춨톶뗘횷ꎺ쯞훝쫐뫓슷71뫅 폊뇠ꎺ234000
뗚튻헂 탷싛 죧맻췸싧짏폐죽냙췲룶뿍뮧ꎬ쓇뻍펦룃폐죽냙췲룶췸싧짌뗪ꆣ ꆪꆪJeff Bezos (Amazonꎮcom 뗄CEO) 뿎쳢뗄퇐뺿놳뺰뫍틢틥 평짌컱늿돶냦뗄ꆶ훐맺뗧ퟓ짌컱놨룦ꆷ쟥컺횸돶ꎺꆰ뗧ퟓ짌컱뿬쯙랢햹쫇쫊펦뿆톧랢햹맛뗄룹놾튪쟳ꎬ쫇쏦뛔뺭볃좫쟲뮯쳴햽ꆢ쳡룟욷없뺺헹솦뗄쫗톡횮[5]슷ꆣꆱ뿉쫇ꎬ퓚쿟뗄뗧ퟓ짌컱욽첨훐ꎬ튻쫇뗧ퟓ짌컱췸햾뗄짌욷탅쾢솿뻞듳ꎬ퓬돉뗧ퟓ짌컱췸햾뿍뮧쯑톰싺ퟣퟔ벺탨튪뗄럾컱뫍닺욷쓑뛈볓듳ꎬ붵뗍뿍뮧샻폃뫍럃컊뗧ퟓ짌컱췸햾뗄탋좤뛈ꎻ뛾쫇뗧ퟓ짌컱췸햾돊쿖룸폃뮧뗄럾컱뫍닺욷[6]탅쾢뗄랽쪽죔좻닉폃뒫춳뗄웕춨췸햾뗄쒣쪽ꎬ쫇ퟮ뺭뗤뗄잧죋튻쏦뗄킧맻ꎬ틲듋떼훂뗧ퟓ짌컱췸햾폃뮧컞램헒떽싺ퟣퟔ짭룶탔뮯뗄쳘헷ꆢ낮뫃뫍탨쟳뗄럾컱닺욷탅쾢ꎬ쯹폐뗄뿍뮧쏦뛔춬튻헅쏦뿗ꆣ 헢훖뗧ퟓ짌컱췸햾뗄릤ퟷ쒣쪽튻뚨믡듸살솽룶훂쏼뗄컊쳢ꎺ튻쫇뗧ퟓ짌컱췸햾좱랦헫뛔탔ꎬ뿍뮧컞램톸쯙웋ힽ룐탋좤뗄럾컱뫍닺욷뗄탅쾢ꎻ뛾쫇뗧ퟓ짌컱췸햾컞램캪뿍뮧뿬쯙믱좡싺ퟣ웤탨튪뗄럾컱뫍닺욷탅쾢쳡릩냯훺ꎬ뿍뮧샫뾪떱잰햾뗣톰헒웤쯼췸햾뗄틢풸늻뛏퓶잿ꆣ짏쫶컊쳢믡룸뗧ퟓ짌컱췸햾듸살훂쏼뗄듲믷ꎬ뗧ퟓ짌컱췸햾쎻폐뗃떽헦헽웕벰짮닣듎뗄풭틲퓚폚뗧ퟓ짌컱췸햾캪뿍뮧쳡릩뗄짌욷톡릺럾컱늻떽캻ꆣ 캪뷢뻶짏쫶컊쳢ꎬ 뇣돶쿖쇋췆볶쾵춳( recommender systems) ꎬ쯼듓뿍뮧뗄탋좤낮뫃돶랢ꎬ췆볶싺ퟣ뿍뮧룶탔뮯탨튪뗄럾컱뫍닺욷ꎬ폖돆룶탔뮯췆볶쾵[1]춳(personalized recommender systems) ꆣ내헕췆볶뛔쿳뗄럖샠뇪ힼꎬ떱잰펦폃ퟮ맣랺뗄췆볶쾵춳폐솽훖샠탍ꎺ튻훖쫇쯑쯷틽쟦쾵춳ꎬ쯼캪뿍뮧췆볶뗄뛔쿳쫇췸튳ꎬ샻폃쫽뻝췚뻲뗄벼쫵뫍랽램ꎬ쿲뿍뮧쳡릩싺ퟣ웤룶탔뮯탨튪뗄췸튳췆볶ꎬ죧baidu뗈ꎻ쇭튻훖쫇뗧ퟓ짌컱췸햾릺컯욽첨쿂ꆢ췆볶뛔쿳쫇컇뫏뿍뮧탨쟳뗄룶탔뮯짌욷ꎬ죧뗧펰ꆢ틴샖뗈ꎬ놻죋쏇돆캪뗧ퟓ짌컱룶탔뮯췆볶쾵춳ꎬ볲돆뗧ퟓ짌컱췆볶쾵춳(recommender system in E - commerce)ꆣ 퓚뗧ퟓ짌컱췆볶쾵춳훐ꎬퟮ훘튪뗄풪쯘쫇룶탔뮯췆볶벼쫵ꎬ쯼쫇움볛췆볶[2]쾵춳탔쓜폅쇓뗄훘튪횸뇪ꎬ쒿잰ퟮ쇷탐뗄췆볶벼쫵죧뇭쯹쪾ꆣ 쯦ퟅ뗧ퟓ짌컱췸햾맦쒣뗄늻뛏삩듳ꎬ럾컱뫍닺욷쫽뻝ꆢ폃뮧탅쾢쫽뻝돉놶퓶뎤ꎬ뛔룶탔뮯췆볶벼쫵쳡돶쇋탂쳴햽ꎬ죧쳡룟룶탔뮯췆볶벼쫵뗄퓋쯣쯙뛈ꎬ볓잿룶탔뮯췆볶쾵춳뗄쪵쪱탔ꆣ쇭튻랽쏦ꎬ떱잰퇐뺿췆볶쯣램뗄죈뗣컊쳢벯훐퓚죧뫎쳡룟췆볶쾵춳뗄ힼ좷탔ꎬ쒿잰룶탔뮯췆볶쾵춳퓚뗧ퟓ짌컱췸햾훐뮹쎻폐맣랺펦폃ꎬ폈웤헫뛔볛룱쿠뛔낺맳뗄럾컱뫍닺욷췆볶ꎬ뗃늻떽뿍뮧뗄죏뿉ꎬ훷튪틲쯘뻍쫇쒿잰뗧ퟓ짌컱췸햾뗄췆볶쾵춳췆볶뗄ힼ좷싊뗍ꎬ폃뮧늻탅죎췆볶쾵춳뗄췆볶뷡맻ꆣ 1
뇭 룷훖췆볶벼쫵 췆볶벼쫵 놳뺰쳵볾 쫤 죫 훷튪늽훨 볆쯣u뗄쿠쯆폃뮧ꎻ춨U뛔I뗄 킭춬맽싋췆볶 U뛔I움럖 맽웤뗃떽i 뗄움럖벶움럖벶뇰 뇰 U 뛔I 뗄 틀뻝U뗄움럖벶뇰ꎬ뗃믹폚쓚죝췆볶 I 뗄쫴탔쳘헷 움럖벶뇰 떽닺욷뗄럖샠웷 믹폚죋샠쳘헷 U뗄죋샠쳘헷뫍뛔I뗄움맘폚U뗄 볆쯣u뗄쿠쯆폃뮧춨맽췆볶 럖벶뇰 죋샠쳘헷 웤뗃떽i뗄움럖벶뇰 쮵쏷U뛔I욫뫃뗄탅죎퓋폃탅죎몯쫽폚룷닺믹폚탅죎췆볶 I뗄쫴탔 몯쫽 욷ꎻ짺돉룷닺욷뗄탲쇐I뗄쫴탔ꆢI싺ퟣ뿍뮧횪쪶볆쯣룷닺욷I뫍뿍뮧탨믹폚횪쪶췆볶뛈 뛔U탨튪뫍탋좤뗄쮵쏷뗄랽쪽 쟳뗄욥엤뛈 볆쯣맘솪맦퓲ꎻ춨맽웤믹폚맦퓲췆볶 U뛔I뗄럃컊믲붻틗볇슼 럃컊붻틗볇슼 믱뗃췆볶 ꎨ힢ꎺU 듺뇭뿍뮧벯ꎻ I 듺뇭닺욷벯ꎻ u 듺뇭믮뚯뿍뮧ꎻ i듺뇭믮뚯닺욷ꆣꎩ 쯤좻킭춬맽싋ퟷ캪튻훖뗤탍뗄췆볶벼쫵폐맣랺뗄펦폃ꎬ쫇뗧ퟓ짌컱췆볶쾵춳훐펦폃ퟮ퓧뫍ퟮ돉릦뗄췆볶벼쫵ꎬ떫뮹튪뛔웤뷸탐늻뛏룄뷸ꆣ킭춬맽싋췆볶벼쫵뗄퇐뺿쫇쒿잰죈뗣컊쳢ꎬ훷튪캧죆킭춬맽싋뗄쾡쫨탔(sparsity) 뫍샤뾪쪼[3ꎬ4]컊쳢(cold - start ) ꎬ떱좻뮹냼삨폃뮧뗚튻듎럃컊ꆢ췆볶쯣램뿉뾿탔뗈컊쳢ꎨ뗚죽헂쿪쾸쏨쫶ꎩꆣ캪뷢뻶킭춬맽싋췆볶쯣램만폐뗄뗤탍컊쳢ꎬힼ좷풤닢폃뮧뗄탋좤ꎬ쳡룟췆볶쾵춳뗄킧싊뫍췆볶훊솿ꎬ뇘탫뛔킭춬맽싋췆볶쯣램볓틔룄뷸ꆣ 킭춬맽싋췆볶쯣램뗄퇐뺿쿖ힴ 헫뛔킭춬맽싋쯣램훐폶떽뗄컊쳢ꎬ톧헟쳡돶쇋쿠펦뗄뒦샭랽램ꎬ웤훐맘힢뛈뷏룟뗄쫇죧뫎뷢뻶폃뮧움럖쫽뻝쾡쫨탔컊쳢ꎬ쒿잰퓋폃ퟮ맣랺뗄뷢뻶냬램폐죧쿂3훖ꎺ (1) 샻폃쒬죏움럖ꆣ 퓚뿍뮧뗄움럖쫽뻝훐ꎬ룸캴움럖쿮쒿뗄럖횵튻룶쒬죏횵믲헟믹폚쿮쒿뗄쿠쯆뛈볆쯣뿍뮧뛔캴움럖쿮뗄럖횵살붵뗍뿍뮧움럖쫽뻝뻘헳뗄쾡쫨뛈ꎬ듓뛸뗃떽쿠쯆폃뮧뗄짺돉훊솿ꎬ떫룃뒦샭쫽뻝쾡쫨탔컊쳢늻뎹뗗ꆣ (2) 샻폃죋릤훇쓜뗄춾뺶ꆣ 샻폃뻛샠ꆢHorting춼ꆢ놴튶쮹췸싧뗈쫖뛎ꎬ쳡룟뿍뮧퓚쿮쒿뻘헳짏붻닦뗄쫽쒿ꎬ퓶볓움럖쫽뻝돭뛈ꆣ룃랽램좷쪵뛔뷢뻶움럖쫽뻝뗄쾡쫨탔컊쳢폐냯훺ꎬ[9]떫붵뗍쇋췆볶쾵춳뗄췆볶ힼ좷뛈ꎬ뛸쟒췆볶쯣램뗄쪵폃탔늻잿ꆣ (3) 샻폃붵캬뗄벼쫵ꆣ 닉폃잱퓚폯틥쯷틽ꆢ웦틬횵럖뷢ꆢ훷돉럖럖컶뗈벼쫵ꎬ냑쾡쫨뗄뿍뮧움럖2
뻘헳뇤돉평훷튪쓚죝ퟩ돉뗄돭쏜뻘헳ꎬ뒦샭쫽뻝쾡쫨컊쳢ꆣ룃랽램뛔뗧ퟓ짌컱췸햾췆볶쾵춳쪵폃탔폐쿔훸뗄쳡룟ꎬ쓜뫜뫃뗘뷢뻶닺욷탅쾢뗄폯틥쿠뷼컊쳢ꎬ떫쫇퓚붵캬훐ꎬ믡뚪쪧튻킩훘튪탅쾢ꎬ붵캬킧맻뮹좡뻶폚뿍뮧움럖쫽뻝뗄쏜벯돌뛈ꎬ죧맻퓚뿕볤쫽쒿캬쫽뫜룟뗄뿍뮧움럖쫽뻝쿂ꎬ뻍컞램놣횤뒦샭뗄킧맻ꆣ 컄쿗[7]훐쫗듎쳡떽킭춬맽싋췆볶벼쫵뗄ꆰ샤웴뚯ꆱ컊쳢ꎬ퓚뫳살뗄퇐뺿훐튲뗃떽쇋탭뛠퇐뺿헟뗄맘힢ꆣ컄쿗[8]죏캪샻폃쿮쒿놾짭쫴탔쳘헷뗄쿠맘탔ꎬ살뷢뻶ꆰ샤웴뚯컊쳢ꆱꆣ 헫뛔삩햹탔컊쳢ꎬ죋쏇쳡돶쇋믹폚쒣탍뗄쯣램ꎬ죧놴튶쮹췸싧쒣탍ꎬ룃랽램뿉틔퓚튻뚨돌뛈짏쳡룟쯣램뗄뿉삩햹탔ꎬ퓚폃뮧탋좤낮뫃뇤뮯뷏짙뗄뮷뺳쿂룃샠쯣램킧맻쏷쿔ꎬ떫통솷쒣탍뗄돉놾뷏룟ꎬ죴쫽뻝룼탂욵랱ꎬ탨훘탂통솷쒣탍ꎬ뛔폚듋샠쾵춳듺볛듳늻첫쫊뫏ꆣ캪뷢뻶짏쫶컊쳢ꎬ컄쿗[10]쳡돶솽훖죧뫎톡퓱쿠쯆폃뮧룄뷸뗄쯣램ꎬ죴폃뮧쫽솿뷏듳쪱ꎬ뿬쯙ꆢힼ좷뗘헒떽쒿뇪뿍뮧뗄ퟮ뷼쇚뻓뇤뗄쿠뛔죝틗ꎬ폐샻폚뷢뻶췆볶쯣램뗄삩햹탔컊쳢ꆣ 뒫춳킭춬맽싋췆볶쾵춳붫폃뮧쿠쯆탔ퟷ캪톡퓱췆볶헟뗄믹ힼꎬ맽뛠뗘틀삵[11]폃뮧쿠쯆탔ꆣ퇮쏷뮨뗈죋쳡돶쇋믹폚뺭퇩뗄킭춬췆볶쯣램ꎬ횸돶췆볶킧맻쫜뛠룶틲쯘펰쿬ꎬ폃뮧뺭퇩쫇톡퓱췆볶헟쪱탨튪뾼싇뗄훘튪틲쯘ꆣ붫폃뮧뺭퇩폫쿠쯆탔쿠뷡뫏ꎬ헻뫏떽뇪ힼ뗄킭춬췆볶뿲볜훐ꎬ쳡룟췆볶뗄훊솿ꆣ 쿫뗃떽췆볶쾵춳뗄ퟮ뫃킧맻ꎬ퇐뺿헟쳡돶믬뫏췆볶벼쫵ꎬ붫킭춬맽싋쾵춳폫쿖폐뗄믹폚쓚죝뗄탅쾢맽싋쾵춳쿠뷡뫏ꆣ킭춬맽싋쾵춳퓚풤닢폃뮧뛔쒳튻쿮쒿뗄룐탋좤돌뛈믲룶탔쳘헷쪱뫜폐ퟷ폃ꎬ떫튻냣늻쓜뫜뫃뗘헒떽몬폐쒳튻쳘뚨쓚죝뗄탅쾢ꎬ틲듋붫킭춬맽싋폫믹폚쓚죝뗄맽싋벼쫵뷡뫏웰살ꆣWebSIFTꆢFabꆢAnatagonomy 뫍RAAP뗈쾵춳뚼쫇닉폃믹폚쓚죝맽싋뫍킭춬맽싋헢솽훖벼쫵ꆣ솽훖췆볶벼쫵쿠뗃틦헃ꎬ죧캪뿋럾킭춬맽싋뗄쾡쫨탔컊쳢ꎬ뿉틔샻폃폃뮧샀맽뗄Web짏풴쓚죝풤닢폃뮧뛔웤쯻풴뗄움볛ꎬ헢퇹뿉틔퓶볓풴움볛뗄쏜뛈ꎬ샻폃헢킩움볛퓙뷸탐킭춬맽싋ꎬ듓뛸쳡룟킭춬맽싋뗄탔쓜ꆣ 놾컄뗄훷튪퇐뺿쓚죝폫ퟩ횯뷡릹 떽쒿잰캪횹ꎬ쯹폐킭춬맽싋ꎨCFꎩ쯣램냑쯼쏇뗄췆볶붨솢퓚튻룶떥튻뗄췆볶ퟩ훐ꎬ캪뷢뻶킭춬맽싋쯣램훐뗤탍뗄컊쳢ꆢ헫뛔뻟폐쏅횪쪶놳뺰뫍붻뮥탔잿뗄쿮쒿췆볶ꎬ쳡룟췆볶쾵춳뗄킧싊폫훊솿ꎬ놾컄쳡돶믹폚쮫탅쾢풴쒣쪽뗄킭춬맽싋쯣램ꎨDual Information Source Model-Based Collaborative Filtering AlgorithmsꎬDISCFꎩꎬ룃랽램에뛏믮뚯폃뮧뛔쒿뇪쿮쒿뗄탋좤돌뛈붨솢퓚솽룶췆볶ퟩꆪꆪ쿠쯆폃뮧췆볶ퟩꎨퟮ뷼쇚뻓벯뫏ꎩ폫볒췆볶ퟩ믹뒡짏ꎬ냑솽룶췆볶ퟩ뗄붨틩뷡뫏웰살ꎬ탎돉뿉뾿뗄탅쾢풴ꎬ좻뫳ꎬ럖컶룷ퟔ펰쿬믮뚯폃뮧뛔쒿뇪쿮쒿뗄좨훘ꎬ볆쯣믮뚯폃뮧뗄ퟮ훕탋좤뛈ꎬ쪵쿖쾵춳췆볶ꆣ 3
놾컄ퟩ횯뷡릹 놾컄릲폐컥헂ꎬ뻟쳥뗄ퟩ횯죧쿂ꎺ 뗚튻헂 탷싛ꎺ볲떥뷩짜쇋놾뿎쳢뗄퇐뺿놳뺰뫍틢틥ꆢ맺쓚췢뗄퇐뺿쿖ힴ틔벰훷튪뗄퇐뺿릤ퟷꆣ 뗚뛾헂 뗧ퟓ짌컱훐췆볶쾵춳퇐뺿ꎺ뷩짜쇋췆볶쾵춳퓚뗧ퟓ짌컱훐ퟷ폃ꎬ룸돶뗧ퟓ짌컱훐뗄췆볶쾵춳뗄쒣탍뫍쫤죫쫽뻝뗄훖샠벰웤쳘헷ꎬ늢뛔췆볶쾵춳뷸탐쇋럖샠틔벰쇐뻙쇋췆볶쾵춳훐쯹펦폃뗄뗤탍벼쫵ꆣ 뗚죽헂 킭춬맽싋췆볶쯣램ꎺ뛔킭춬맽싋췆볶쯣램뗄릤ퟷ풭샭ꆢ쫤죫쫤돶뫍폅뗣뷸탐쇋뷩짜ꎬ늢훘뗣뷩짜ퟮ쇷탐뗄킭춬맽싋쯣램뗄럖샠ꎬ뛔쿖폐뗄킭춬맽싋쯣램훐듦퓚뗄컊쳢뷸탐쇋퇐뺿ꎬ럖컶떱잰뷢뻶랽램뗄폅좱뗣ꆣ 뗚쯄헂 믹폚쮫탅쾢풴뗄킭춬맽싋쯣램퇐뺿ꎺ쳡돶쇋튻훖믹폚쮫탅쾢풴뗄킭춬맽싋쯣램ꎬ룸돶쇋룃랽램뗄쮼슷틔벰뻟쳥쪵쿖늽훨ꎬ늢춨맽쫔퇩럂헦닢쫔볬퇩쇋탂랽램뗄췆볶킧맻ꆣ 뗚컥헂 ퟜ뷡폫햹췻ꎺ뛔놾컄뗄훷튪퇐뺿릤ퟷ뷸탐ퟜ뷡ꎬ늢룸돶뷸튻늽뗄릤ퟷ쒿뇪폫볆뮮ꆣ 4
뗚뛾헂 뗧ퟓ짌컱췸햾훐췆볶쾵춳퇐뺿 뗧ퟓ짌컱췸햾훐췆볶쾵춳펦폃쿖ힴ 뗧ퟓ짌컱췸햾훐췆볶쾵춳ퟷ폃 쿖퓚ꎬ룼뛠뗄죋뷨훺췸싧살톡퓱ퟔ벺룐탋좤뗄짌욷ꎬ떫쫇ꎬ쯦ퟅ췸햾짏맋[13]뿍ꆢ짌욷폫붻틗뗄돉놶퓶뎤ꎬ뗧ퟓ짌컱췸햾춬퇹돶쿖쇋탅쾢맽퓘컊쳢ꆣ캪쓜룸뗧ퟓ짌컱췸햾뿍뮧쳡릩룼볓룶탔뮯뗄럾컱ꎬ룶탔뮯췆볶쾵춳퓚뗧ퟓ짌컱췸햾훐냧퇝풽살풽훘튪뗄뷇즫ꆣ췆볶쾵춳쓜쪵쿖췸햾ퟔ뚯뮯ꆢ룶탔뮯럾컱ꎬ캪쎿룶맋뿍쳡릩튻룶늻춬뗄럾컱뮷뺳ꆣ [14]췆볶쾵춳뿉틔듓죽룶랽쏦쳡룟뗧ퟓ짌컱췸햾뗄쿺쫛뛮ꎺ ꋅ 샀헟뇤돉릺싲헟ꎨBrowsers into buyersꎩ 튻룶췸햾뗄샀헟뺭뎣맢맋췸햾ꎬ떫쫇뫜짙ퟶ돶릺싲뻶뚨ꆣ췆볶쾵춳뿉틔냯훺샀헟랢쿖쯻룐탋좤뗄ꆢ쾣췻릺싲뗄짌욷ꆣ ꋆ 쪵쿖붻닦쿺쫛ꎨCross-sellꎩ 췆볶쾵춳캪릺싲헟쳡릩뛮췢뗄짌욷탅쾢ꎬ살쪵쿖붻닦쿺쫛ꆣ튻룶돶즫뗄췆볶쾵춳ꎬ믡쳡룟뗧ퟓ짌컱췸햾뗄뚩떥ꆣ샽죧ꎺ짌욷쿺쫛퓚좷죏뗄맽돌훐ꎬ췸햾믡룸쿻럑헟쳡릩틑릺짌욷뗄맘솪탔짌욷ꆣ ꋇ 쳡룟뿍뮧훒돏뛈ꎨLoyaltyꎩ 퓚뗧ퟓ짌컱뗄쫀뷧샯ꎬ뺺헹뿉쓜뻍쫇떥믷ꆢ샀뗈볲떥뗄쫳뇪닙ퟷꎬ펮뗃맋뿍뗄훒돏뛈닅쫇짌튵훐ퟮ훘튪뗄닟싔ꆣ췆볶쾵춳퓚뗧ퟓ짌컱췸햾폫쿻럑헟횮볤뒴붨뢽볓뗄볛횵맘쾵살쳡룟쿻럑헟뗄훒돏뛈ꆣ뗧ퟓ짌컱췸햾춶죫뺫솦죏쪶ꆢ쇋뷢폃뮧ꎬ췆볶쾵춳뛔웤뷡맻뢶훮쪵쪩ꎬ캪폃뮧쳡릩튻룶쓜싺ퟣ웤탨튪뗄룶탔뮯붻뮥뷧쏦ꎬퟷ캪믘놨ꎬ폃뮧붫믡릺싲췸햾쳡릩뗄쓜싺ퟣ탨튪뗄짌욷ꆣ헢퇹ꎬ폃뮧풽살풽틀뾿췆볶쾵춳ꎬ룦쯟췆볶쾵춳쯻뗄낮뫃ꎬ뛔췸햾뗄훒돏뛈튲믡풽살풽룟ꆣ벴쪹뺺헹헟뻟놸쇋ퟶ춬퇹릤ퟷ뗄쓜솦ꎬ쿻럑헟뇘탫튪뢶돶룼뛠뗄쪱볤뫍뺫솦룦쯟뺺헹헟쓣틑뺭헆컕뗄맘폚쿻럑헟뗄훖훖탅쾢ꆣퟮ뫳ꎬ퓚쿻럑헟ퟔ짭횮볤뒴붨맘쾵춬퇹뿉틔쳡룟훒돏뛈ꆣ쿻럑헟샖폚폫췸햾룸쯻췆볶뗄폃뮧붻쇷ꎬ훰붥뗘쯻뻍믡쾲뮶헢룶췸햾ꆣ 뗧ퟓ짌컱췸햾훐췆볶쾵춳펦폃뻙샽 뗧ퟓ짌컱뗄뿬쯙랢햹췆뚯췆볶쾵춳뗄랢햹ꆣ맺췢탭뛠듳탍뗄뗧ퟓ짌컱췸햾ꎬ샽죧ꎺAmazonꎮcom(wwwꎮamazonꎮcom)ꎬMy CDNOW(wwwꎮcdnowꎮcom)ꎬeBay(wwwꎮebayꎮcom)ꎬMoviefinderꎮcom(wwwꎮmoviefinderꎮcom)뗈뗈쪹폃쇋룷훖탎쪽뗄췆볶쾵춳ꆣ맺쓚튲폐뫜뛠웳튵샻폃췆볶쾵춳쳡룟ퟔ짭뗧ퟓ짌컱췸햾욷없뺺헹솦뗄틢쪶늻뛏퓶잿ꆣ죧탂샋퓚쿟릺컯ꆢ쯑뫼짌돇뫍쳔놦췸햾뚼폐쓜릻싺ퟣ췸햾뿍뮧룶탔뮯탨튪뗄췆볶쾵춳ꆣ뇭훐ퟜ뷡쇋짏쫶쳡떽맺췢뗄뗧5
ퟓ짌컱췸햾췆볶쾵춳펦폃쪵샽ꆣ 뇭 췆볶쾵춳펦폃뻙샽 뗧ퟓ짌컱췸햾쏻돆 췆볶쾵춳쫽뻝 췆볶쾵춳벼쫵 췆볶뷡맻뗄탎쪽 Amazonꎮcom Item to Item Correlation Customers who BoughtSimilar Item Organic Navigation Purchase data Eyes Email AttributeBasedKeywords/freeform Amazonꎮcom DeliversEmail Attribute Based Selection options People to People Correlation Book Matcher Top N List Request List Likert Average RatingCustomer Comments Aggregated Rating Likert Text Organic Navigation Text CommentsCDNOW Similar Item Item to Item Correlation Organic Navigation Album Advisor Top N List Purchase data Keywords/freeform Organic Navigation My CDNOW People toTop N List People Correlation Likert Request List eBay Average RatingFeedback Profile Aggregated Rating Likert Text Organic Navigation Text CommentsMoviefinderꎮcom Item to Item Correlation Match Maker Similar Item Navigate to an item Editor’s choice Top N List People to People Correlation Keywords/freeform Ordered Search We Predict Aggregated Rating Selection options Results Likert Organic Navigation Average Rating 뗧ퟓ짌컱훐췆볶쾵춳쒣탍 뗧ퟓ짌컱훐췆볶쾵춳쒣탍 췆볶쾵춳릤ퟷ쇷돌쫇ꎺ맋뿍ꆢ닺욷폫좺쳥뗄쿠맘쫴탔ퟷ캪쫤죫쫽뻝ꎬ샻폃튻쾵쇐뗄췆볶랽램헒떽싺ퟣ쒿뇪폃뮧룶탔뮯탨쟳뗄췆볶뷡맻ꎬ췹췹믡틔붨틩ꆢ풤닢움럖뫍움싛뗄탎쪽돶쿖ꆣ펦폃돌탲냑뷡맻틔튻훖룶탔뮯뗄짌욷탎쪽뒫뗝룸쒿뇪폃뮧ꆣퟮ뫳튻늽뻍쫇쾵춳웋ힽ폃뮧뗄랴삡틢볻ꎬ캪쿂튻듎뗄췆볶ퟶ뫃ힼ놸ꆣ 컄쿗[14]냑췆볶쾵춳럖돉죽룶훷튪뗄늿럖ꎺI/O릦쓜ꆢ췆볶랽램폫짨볆벼쫵ꎨ붻뮥뷧쏦짨볆ꆢ뷡맻뇭쿖탎쪽ꆢ뒫뗝랽램뗈ꎩꆣ떱좻ꎬ헢죽룶릦쓜쒣뿩늻쫇6
뛀솢퓋탐ꎬ쯼쏇쿠뮥펰쿬폫훆풼ꆣ샽죧ꎺ튻훖췆볶랽램탨튪쳘쫢뗄쫤죫쫽뻝ퟷ캪믹뒡ꎬ췆볶뷡맻뗄탎쪽튲쫇좷뚨뗄ꎬ짨볆벼쫵쯼쏇뗄뇤뮯ퟶ돶쿠펦뗄뗷헻ꆣ뗧ퟓ짌컱훐췆볶쾵춳쒣탍ꎬ죧춼쯹쪾ꎺ 췆볶랽램 풭쪼쫽뻝쳡좡 쫖릤럖샠 쒿뇪폃뮧쫤죫좺쳥쫤죫 춳볆럖컶 틾탎쫤죫쿮쒿쫴탔 Attribute—based 쿔탔쫤죫릺싲샺쪷볇슼 Item—to—Item correlation 맘볼ퟖ움럖 /쿮쒿 쫴탔힢뷢 user—to—user correlation 릺싲샺쪷볇슼삩햹쿮쒿 쫤돶 붨틩 풤닢움럖 움싛 뗧ퟓ짌컱췸햾췆볶짌욷뷓뿚 랴삡 랴삡 춼 뗧ퟓ짌컱훐췆볶쾵춳쒣탍 췆볶쾵춳뗄쫤죫쫽뻝 Resnick퓚컄쿗[15]훐횸돶ꎬ듓뗚튻룶췆볶쾵춳뾪쪼ꎬ죧뫎쏨쫶폃뮧뗄탋좤쒣탍ꎬ돉캪늻춬췆볶벼쫵탨튪뿋럾뗄쫗튪컊쳢ꆣ쇋뷢폃뮧뗄탋좤ꎬ쫇듓쫤죫쫽뻝뾪쪼뗄ꎬ샽죧춨맽폃뮧뛔췸튳뗄럃컊췚뻲ꎬ쫇닺짺폃뮧탋좤쒣탍뎣폃뗄랽램ꆣ쒿잰뇈뷏돉릦뗄췆볶쾵춳ꎬ샽죧ꎺMovielensꎬ FabꎬEntree restaurant recommender뗈뗈ꎬ뚼듓폃뮧쫤죫쫽뻝훐쳡좡룷훖룷퇹뗄폐폃탅쾢ꆣ춨뎣퇐뺿컄쿗냑쫤죫쫽뻝럖돉솽훖샠탍ꎺ폃뮧쫽뻝뫍짌욷쫽뻝ꆣ 7
폃뮧쫽뻝폖폐쯄룶늻춬뗄랽쏦ퟩ돉ꎺ춳볆쫽뻝ꆢ폃뮧움럖쫽뻝ꆢ폃뮧쒣탍쫽뻝폫폃뮧붻틗쫽뻝ꆣ퓚퓧웚뗄폃뮧쫽뻝뗄퇐뺿훐ꎬ횻헫뛔춳볆쫽뻝뫍폃뮧뛔룐탋좤짌욷뗄움럖쫽뻝ꎬ쿱Movielens췆볶쾵춳뻍쫇샻폃헢솽훖쫽뻝췪돉췆볶ꆣퟮ뷼ꎬ탭뛠퇐뺿헟샻폃web췚뻲뫍웤쯼췸싧벼쫵ꎬ쳡좡폃뮧퓚샀췸튳뗄잱퓚탐캪ꎬ쳡솶돉폃뮧뗄탋좤쒣탍ꆣ붻틗쫽뻝쫇횸폃뮧릺싲짌욷뗄탅쾢믡뒢듦퓚붻틗쫽뻝뿢훐ꆣ짌욷쫽뻝횸뗄쫇췆볶짌욷뗄훷튪쫴탔ꆣ샽죧ꎺ뛔폚뗧펰뗄췆볶쾵춳Movielens훐ꎬ짌욷쫽뻝횸뗄뗧펰뗄훷쳢ꆢ떼퇝ꆢ퇝풱ꆢ랢탐쪱볤뗈뗈ꆣ뛔폚탂컅ꆢ컄쿗뗄췆볶쾵춳Fab닺욷쫽뻝쫇탂컅뗄훷쳢ꆢ탂컅뗄맘볼ퟖ뗈뗈ꆣ 뇭뗧ퟓ짌컱췆볶쾵춳훐쫤죫쫽뻝샠탍 쫽뻝샠탍 쮵쏷 폃뮧뗄 탕쏻ꎬ 쓪쇤ꎬ 탔뇰ꎬ 횰튵ꎬ 돶짺죕웚ꎬ 솪쾵랽쪽ꎬ 힡횷ꎬ 탋좤낮뫃ꎬ쫴탔쳘헷 킽돪ꎬ 뷌폽뺭샺뗈뗈ꆣ 움럖뗈벶ꎬ 죧ꎺ샫즢쫽횵뫍솬탸뗄쫽횵ꎻ믲헟룐쟩즫닊쏨쫶ꎺ폅탣ꎬ 뫃ꎬ움럖쫽뻝 늻뫃ꎬ 뫜닮뗈뗈ꆣ 돖탸샀ꎬ 떥믷쪱볤ꎬ듲뾪췸햾뗄솴뷓ꎬ 놣듦ꎬ 듲펡ꎬ 맶뚯ꎬ즾돽ꎬ 듲폃뮧탐캪 뾪ꎬ 맘뇕ꎬ 쮢탂튳쏦ꎬ톡퓱ꎬ 뇠벭ꎬ 쯑쯷ꎬ 뢴훆ꎬ 햳쳹ꎬ쫕님뫍쿂퓘쒣쪽쫽뻝 튳쏦쓚죝뗈뗈ꆣ 붻틗쫽뻝 붻틗죕웚ꎬ 붻틗쫽솿ꎬ볛룱ꎬ 헛뿛뗈뗈ꆣ 죧뛔뗧펰ꎺ훷튪퇝풱ꎬ 훷쳢ꎬ 랢탐쪱볤ꎬ 볛룱ꎬ 랢탐짌뗈뗈ꎻ 죧뛔췸닺욷쫽뻝 햾ꎺ맘볼듊ꎬ 솴뷓ꎬ 샀듎쫽ꎬ 훷쳢뗈뗈ꎮ 돽쇋짏쫶뗄뎣폃뗄쫤죫쫽뻝탎쪽횮췢ꎬ튻킩퇐뺿헟쳡돶쇋뢴퓓뗄쏨쫶쫤죫쫽뻝뗄랽램ꆣShapira퓚2006쓪쳡돶뗄샀맽돌훐쫳뇪뗄퓋뚯뫍맶뚯쪱볤ꆢ폃[16]뮧샀췸햾뗄쪱볤뇪ힼ뮯캪쯹샀튳쏦뗄듳킡뫍튳쏦뗄뎬솴뷓뗄쫽쒿ꆣ쯹폐뗄쫤죫쫽뻝죧뇭쯹쪾ꆣ떫쫇ꎬ퓚뇭훐늢쎻폐쳡떽좺쳥쫽뻝뗄쫤죫ꎬ놾컄뗚쯄헂쓚죝뻍쫇틽죫좺쳥볒쫽뻝ꎬ살쳡룟췆볶뗄훊솿ꆣ 췆볶쾵춳뗄럖샠 내헕췆볶쾵춳캪쒿뇪폃뮧쳡릩뗄췆볶뷡맻뗄ퟔ뚯뮯돌뛈뫍돖뻃탔돌뛈뇪ힼ[17]뛔뗧ퟓ짌컱췸햾훐뗄췆볶쾵춳뷸탐럖샠ꎺ ꎨ1ꎩퟔ뚯뮯돌뛈ꎨdegree of automationꎩ 쫖릤췆볶ꆢ냫ퟔ뚯췆볶뮹쫇췪좫뗄ퟔ뚯췆볶듺뇭췆볶쾵춳뗄ퟔ뚯뮯돌뛈뗄룟뗍ꎬ췆볶쾵춳캪쒿뇪폃뮧쳡릩싺ퟣ웤탨튪뗄룶탔뮯럾컱맽돌훐ꎬ탨튪쒿뇪폃뮧붻뮥풽짙ꎬ쾵춳뗄ퟔ뚯뮯돌뛈풽룟ꆣ ꎨ2ꎩ돖뻃탔돌뛈ꎨdegree of persistenceꎩ 쮲볤췆볶ꆢ폐랴펦쪱볤췆볶떽돖뻃탔췆볶쫇췆볶쾵춳돖뻃탔뗄뇤뮯랶캧ꆣ8
럖컶쒿뇪폃뮧헽퓚닙ퟷ뗄떥튻믡뮰ꎨSessionꎩꎬ늻뫍떱잰폃뮧맽좥뗄죎뫎탅쾢붨솢맘솪뗄췆볶돆캪쮲볤췆볶ꎻ랴횮ꎬ뗧ퟓ짌컱췸햾뗄췆볶쾵춳샻폃쒿뇪폃뮧뗄힢닡탅쾢뫍맽좥뗄샀릺싲탐캪뛸뗃떽뗄췆볶쫇돖뻃탔췆볶ꆣ 짏쫶뇪ힼ쫇펦폃뷏캪맣랺뗄췆볶쾵춳뗄럖샠랽램ꎬ돽듋횮췢ꎬ쒿잰ꎬ룶탔뮯돌뛈ꎨdegree of personalizationꎩ놻퇐뺿헟폃살움볛뗧ퟓ췸햾췆볶쾵춳탔쓜폅쇓뗄훘튪횸뇪ꎬ쯼뇭쪾췆볶뷡맻뫍쒿뇪폃뮧룶탔뮯탋좤쳘헷뗄욥엤돌뛈ꆣ 내헕룶탔뮯돌뛈뗄뇪ힼ냑췆볶쾵춳뮮럖캪쿂쏦쯄훖샠탍ꎺ ꎨ1ꎩ컞룶탔뮯뗄췆볶쾵춳ꎨNon-Personalized Recommendationsꎩ 믹폚웤쯻폃뮧뛔짌욷뗄욽뻹움럖ꎬ쿲믮뚯뿍뮧닺짺췆볶뷡맻ꆣ췆볶쾵춳폫믮뚯뿍뮧쫇췪좫뛀솢뗄ꎬ쯹틔쯼캪쯹폐뗄뿍뮧췆볶뗄뷡맻췪좫튻퇹ꆣ컞룶탔뮯뗄췆볶쾵춳쫇ퟔ뚯ꎬ틲캪쯼뫜짙탨튪뿍뮧뗄탅쾢쫤죫ꎻ늻믡뇦죏믮뚯뿍뮧뗄튻룶믡뮰떽쇭튻룶믡뮰ꎬ폫뿍뮧쫇췑샫뗄ꎬ쯼폖쫇쮲볤뗄ꆣ ꎨ2ꎩ믹폚닺욷쳘헷뗄췆볶쾵춳ꎨAttribute-Based Recommendationsꎩ 캪믮뚯뿍뮧쳡릩췆볶뷡맻ꎬ믹폚닺욷뗄쫴탔쳘헷ꆣ룃쾵춳퓚릤ퟷ맽돌훐탨튪믮뚯뿍뮧쿔쪽뗘쫤죫쯹탨튪뗄짌욷뗄쫴탔쳘헷ꎬ틲듋쫴폚쫖릤췆볶랽쪽ꆣ죧맻듓돖뻃탔뷇뛈뿉틔쫇쮲쪱ꎬ튲뿉틔쫇돖뻃뗄ꎬ맘볼퓚폚뗧ퟓ짌컱췸햾췆볶쾵춳쫇럱뒦샭믮뚯뿍뮧뗄탅쾢뫍샀탐캪볇슼ꆣ ꎨ3ꎩ믹폚닺욷맘솪뗄췆볶쾵춳ꎨItem-to-Item Correlation Recommendationsꎩ 믹폚믮뚯뿍뮧룐탋좤뗄짌욷ꎬ췆볶쾵춳췆볶뫍룃짌욷맘솪탔잿뗄짌욷ꎬ죧맻췆볶쾵춳죏캪믮뚯뿍뮧뗄릺싲탐캪늻믡랢짺룄뇤ꎬ뻍쫇ퟔ뚯뗄췆볶랽쪽ꎻ죧맻믮뚯뿍뮧퓚췆볶뗄맽돌훐쿔쪽뗘쫤죫룐탋좤뗄짌욷ꎬ튲뿉틔죏캪쫇쫖릤랽쪽ꆣ떫쿠맘탔짌욷췆볶쾵춳춨뎣뚼쫇쮲쪱뗄ꎬ틲캪늻탨튪믮뚯뿍뮧뗄릺싲짌욷뗄샺쪷볇슼ꎬ횻탨쇋뷢믮뚯뿍뮧떱잰톡퓱뗄짌욷ꆣ ꎨ4ꎩ쿠맘탔폃뮧췆볶쾵춳ꎨPeople-to-People Correlation Recommendationsꎩ 풴폚탅쾢맽싋쾵춳훐뷨훺폃뮧ퟩ뗄맛뗣캪믮뚯폃뮧췆볶쿮쒿ꎬ폖돆캪킭춬맽싋쾵춳ꆣ뷨훺폚퓚뗧ퟓ짌컱췸햾훐틑뺭릺싲맽짌욷뗄폃뮧폫믮뚯폃뮧뗄쿠쯆탔뷸탐췆볶ꆣ쾵춳쫗쿈럖컶믮뚯폃뮧뗄탋좤쒣탍ꎬ퓚폃뮧좺훐볆쯣폫쯻폐쿠쯆탋좤쳘헷뗄ퟮ뷼쇚뻓벯ꎻퟛ뫏ퟮ뷼쇚뻓벯훐폃뮧뛔쒿뇪쿮쒿뗄움볛ꎬ살볆쯣돶믮뚯폃뮧뛔쒿뇪쿮쒿쾲낮돌뛈뗄풤닢횵ꆣ룃쾵춳늻탨튪믮뚯뿍뮧쫤죫탅쾢ꎬ룹뻝믮뚯폃뮧떱잰뗄탐캪살쪵쿖췆볶ꎬ뛸쟒늻춬뗄폃뮧룹뻝웤룶죋쳘헷뗃떽룷늻쿠춬뗄췆볶뷡맻ꎬ룶탔뮯돌뛈뫜룟ꆣ믹폚닺욷쫴탔뗄췆볶쾵춳뫍닺욷쿠맘탔췆볶쾵춳탨튪뛔닺욷쓚죝쳘헷뗄볆쯣ꎬ볆쯣쓑뛈듳뛸쟒늻틗랢쿖믮뚯폃뮧탂뗄탋좤뗣ꆣ틲듋ꎬ킭춬맽싋벼쫵쫇뗧ퟓ짌컱췸햾췆볶쾵춳훐펦폃ퟮ맣랺튲쫇ퟮ돉릦뗄벼쫵ꆣ 9
뗧ퟓ짌컱췸햾훐췆볶쾵춳뗤탍벼쫵뷩짜 췆볶쾵춳릤ퟷ럖죽늽ꎺ쫗쿈ꎬ뷨훺폃뮧뗄쫤죫쫽뻝ꎬ췚뻲돶폃뮧뗄탋좤쒣탍ꎻ좻뫳ꎬ뷨훺췆볶쯣램볆쯣돶췆볶뷡맻ꎬퟮ뫳냑췆볶뷡맻돊쿖룸폃뮧ꆣ볆쯣췆볶뷡맻맽돌훐닉폃뗄벼쫵쫇럇뎣훘튪뗄ꎬ쯼쫇췆볶쾵춳돉냜뗄맘볼ꆣ쒿잰놻뻸듳늿럖퇐뺿헟뷓쫜뗄췆볶벼쫵폐죽훖ꎺ킭춬맽싋ꎨcollaborative filteringꎩꆢ믹폚쓚죝뗄맽싋ꎨcontend-based filtering ꎩ뫍믬뫏췆볶랽램ꎨhybrid approachꎩꆣ 킭춬맽싋벼쫵ꎨCollaborative filtering approachꎩ 킭춬맽싋벼쫵쫇펦폃ퟮ퓧뫍ퟮ돉릦뗄췆볶벼쫵ꎬ쯼뷨훺폃뮧쫽뻝뿢ꎬ퓋폃킭춬맽싋쯣램췪돉뛔믮뚯뿍뮧뗄췆볶ꆣ킭춬맽싋벼쫵훐ퟮ맘볼뗄튻늽쫇헒떽폫믮뚯뿍뮧폐ퟅ쿠쯆탋좤낮뫃뗄ퟮ뷼쇚뻓벯ꆣ좻뫳ꎬ춨맽ퟮ뷼쇚뻓벯뗄낮뫃캪믮[18]뚯뿍뮧쪵쿖췆볶ꆣ컒쏇폖냑킭춬맽싋럖돉솽룶샠탍ꎺ웴랢쪽ꎨheuristic-based methodꎩ킭춬맽싋벼쫵뫍믹폚쒣쪽뗄ꎨmodel-based methodꎩ킭춬맽싋벼쫵ꆣ 웴랢쪽킭춬맽싋벼쫵ꎨHeuristic-based collaborative filteringꎩ샻폃믮뚯뿍뮧뗄움럖쫽뻝ꆢ샀췸햾뗄쪱볤ꆢ폃뮧붻틗쫽뻝뫍웤쯼쿠맘뗄틾탎뗄쫽뻝ퟷ캪췆볶쾵춳뗄쫤죫쫽뻝ꎬ퓚헻룶쿻럑헟쫽뻝뿢훐살볆쯣췆볶뷡맻ꆣ뒫춳뗄웴랢쪽킭춬맽싋벼쫵춨뎣퓚쯹폐뗄쿻럑헟쫽뻝뿢훐춨맽쿠쯆탔볆쯣ꎬ헒떽믮뚯뿍뮧뗄쿠쯆폃뮧ꆣ춨맽ퟮ뷼쇚뻓벯ꎨKNNꎩ훐쿠쯆폃뮧뛔믮뚯뿍뮧캴움럖맽뗄쿮쒿뗄럖쫽ꎬ볆쯣돶믮뚯뿍뮧뛔룃쿮쒿뗄움럖ꎬ좻뫳내뗃럖뗄뛠짙살ퟩ돉췆볶뗄쿮쒿뗄벯뫏ꎬ쿲믮뚯뿍뮧췪돉췆볶ꆣ 믹폚쒣쪽뗄킭춬맽싋벼쫵ꎨModel-based collaborative filtering methodꎩ퓚통솷쫽뻝뗄믹뒡짏ꎬ샻폃튻뚨뗄벼쫵ꎬ쫗쿈닺짺튻룶쒣쪽ꆣ좻뫳ꎬ폃쪵퇩쫽뻝살볬닢쒣쪽뗄헽좷탔ꆣ냑믮뚯뿍뮧뗄쫽뻝쫤죫떽쒣쪽훐ꎬ뗃돶췆볶쇐뇭믲헟볆쯣돶믮뚯뿍뮧뛔캴움럖쿮쒿뗄움럖ꆣ솽룶쯣램뗄쟸뇰퓚폚ꎺ잰헟샻폃릫쪽캪쎿튻룶폃뮧볆쯣움럖ꎬ뫳튻룶폃믮뚯폃뮧뗄쫽뻝쫤죫떽쒣쪽훐살뗃떽췆볶쇐뇭뫍췪돉ퟮ뫳뗄췆볶ꆣ 믹폚쓚죝맽싋벼쫵ꎨContent-based filtering approachꎩ 뷨훺탅쾢볬쯷벼쫵뫍탅쾢맽싋랽램ꎬ럖컶닺욷쫴탔쳘헷뗄쓚죝ꎬ펦폃쇚뻓몯쫽뫍럖샠벼쫵럖컶뫍뻛샠닺욷ꎬ췆볶쾵춳뇈뷏폃뮧뗄탅쾢떵낸뫍뻛샠뫳뗄닺[19]욷쳘헷닺짺췆볶뷡맻ꆣ 룹뻝췸햾뿍뮧뗄샀쓚죝캪웤릹퓬컄떵쒣탍뫍붨솢췸햾쓚죝ꎨ죧뗧ퟓ짌컱췸햾훐뗄짌욷ꎩ뗄럖샠쒣탍ꎬ쫇룃랽램돉냜폫럱뗄맘볼틲쯘ꆣ쒿뇪뿍뮧샀췸햾뗄맽돌훐ꎬ틀뻝쒿뇪뿍뮧헽퓚샀쓚죝뗄컄떵쒣탍ꎬ췸햾닩헒췆볶폫룃컄떵쒣탍욥엤믲쿠쯆뗄닺욷ꆣ 쯣램볲떥ꆢ췆볶킧맻폐킧쫇룃랽램뗄뎤뒦ꆣ늻ퟣ횮뒦퓚폚쎻폐맘힢쒿뇪뿍10
뮧탋좤뗄잨틆탔ꎬ횻쓜캪쒿뇪뿍뮧췆볶틑듦퓚탋좤쿠쯆뗄닺욷뫍럾컱ꎬ뛔닺욷뫍럾컱뗄럧룱뫍욷훊쟸럖뛈킡ꆣ췆볶쾵춳샻폃룃랽램탨튪ퟶ뫜뛠믹뒡탔릤ퟷꎬ죧뛔쒿뇪뿍뮧샀뗄쓚죝붨쒣ꎬ욥엤췸햾훐폫횮맘솪뛈룟뗄닺욷쓚죝ꎬ믡펦폃떽췸튳쳘헷쳡좡벼쫵늢럖컶췸햾붨쒣뗄폯퇔쳘뗣뗈ꆣ쒿잰췸튳뛠쫇럇뷡릹뮯믲냫뷡릹뮯뷡릹ꎬ쳡좡췸튳쓚죝뗄쳘헷쿲솿뫍쯼쏇횮볤쿠쯆탔볆쯣쫇튻룶쓑쳢ꆣ쯦ퟅ췸튳뫍쒿뇪뿍뮧쫽솿뗄뿬쯙퓶뎤ꎬ믹폚맽싋쓚죝쯣램쏦쇙룼듳뗄쳴햽ꆣ 믬뫏뗄췆볶쾵춳벼쫵ꎨHybrid filtering approachꎩ 캪뿋럾쿖듦뗄벼쫵랽램뗄뻖쿞탔ꎬ퇐뺿헟쳡돶냑짏쫶솽훖랽램폐믺뷡뫏웰살ꎬ쳡룟췆볶쾵춳훊솿ꎬ돆횮캪믬뫏뗄췆볶벼쫵ꎨHybrid filtering approachꎩꆣ 믬뫏췆볶벼쫵뿉틔럖돉죽훖샠탍ꎺꎨ1ꎩ냑튻훖랽램뗄풪쯘뫍쳘탔틽죫떽쇭췢튻훖랽램훐좥ꎻ샽죧ꎺMelville퓚컄쿗[20]훐냑믹폚쓚죝뗄풤닢웷틽죫떽킭춬맽싋쾵춳훐ꎬ좥볆쯣캴움럖쿮쒿뗄풤닢횵ꎬ살뷢뻶쫽뻝뗄쾡쫨탔컊쳢ꆣꎨ2ꎩ냑솽훖랽램뗄췆볶뷡맻뷡뫏웰살뾼싇ꆣ샽죧ꎺClaypool퓚컄쿗[21]훐쳡돶냑킭춬맽싋뗄췆볶뷡맻폫믹폚쓚죝뗄맽싋볲떥뗄쿟탔뷡뫏ꆣꎨ3ꎩ틽죫웤쯼탅쾢뗄믹뒡짏쳡돶뛀쳘뗄폃뮧쒣탍ꆣ샽죧ꎺ틽죫뛔쿠쯆폃뮧뗄탅죎뛈ꎨtrustꎩ컊쳢뗈뗈ꆣ믬뫏췆볶쾵춳뗄럖샠틔벰펦폃뗄벼쫵퓚뇭훐뷸탐쇋ퟜ뷡ꎺ 뇭 믬뫏췆볶쾵춳벼쫵 Hybrid Techniques Typical papersMooney & Royꎬ 1999 Feature Bayesian Condliffꎬ et alꎬ 1999 combining ClusteringKimꎬ et alꎬ 2006 Recommendation Linear Claypoolꎬ et alꎬ 1999 result combining combination Probabilistic Popesculꎬ et alꎬ 2001 Comprehensive and unique model approach Maximum Jinꎬ et alꎬ 2005 entropy 놾헂킡뷡 놾헂훷튪뷩짜쇋췆볶쾵춳뛔뗧ퟓ짌컱췸햾뗄ퟷ폃틔벰펦폃쿖ힴ뗄뻙샽ꆢ룸돶췆볶쾵춳뗄튻냣쒣탍늢럖컶쇋췆볶쾵춳훐뗄쫤죫쫽뻝ꆣ훘뗣뷩짜쇋췆볶쾵춳뗄럖샠뇪ힼ벰웤뗄샠탍뫍쪵쿖췆볶쾵춳훐뎣폃뗄벸훖뗤탍뗄벼쫵랽램ꎬ냼삨킭춬맽싋뗄벼쫵랽램ꆢ믹폚쓚죝맽싋랽램뫍믬뫏췆볶쾵춳벼쫵ꆣ 11
뗚죽헂 킭춬맽싋췆볶쯣램 킭춬맽싋췆볶쯣램 킭춬맽싋췆볶쯣램릤ퟷ풭샭 퓚쪵볊짺믮훐ꎬ뛔폚늻쫬쾤뗄컊쳢믲쫂컯ꎬ죋쏇췹췹튪톯ퟔ벺뗄엳폑믲쫇탅죎뗄죋ꎬ룹뻝쯻쏇뗄에뛏뫍뾴램살ퟷ돶ퟔ벺뗄톡퓱ꆣ [23]킭춬맽싋췆볶쯣램뗄쮼슷쫇ꎺ죴폃뮧횮볤폐쿠뷼탋좤ꎬ쯻쏇뿉쓜믡뛔춬퇹뗄닺욷룐탋좤ꆣ틲듋ꎬ췆볶쾵춳볇슼ꆢ헻샭쏨쫶폃뮧탋좤쳘헷뗄쫽뻝ꎬ듓훐볆쯣뗃돶뻟폐쿠쯆탋좤낮뫃뗄폃뮧ꎨ쇚뻓폃뮧ꎩꎬ룹뻝쇚뻓폃뮧뛔췆볶닺욷뗄틢볻살움뚨쒿뇪폃뮧뗄탋좤뛈ꆣ쇭튻훖쮼슷쫇ꎺ폃뮧뿉쓜뛔폐맘솪믲쿠쯆뗄짌욷룐탋좤ꎬ볆쯣짌욷횮볤뗄쿠쯆돌뛈ꎬ캪폃뮧췆볶폫쿠쯆믲맘솪뗄짌욷ꆣ잰튻훖쮼슷믹폚폃뮧폫폃뮧횮볤뗄맘쾵ꎬ뛸뫳튻훖쮼슷퓲ퟅ퇛폚짌욷폫짌욷횮볤뗄맘솪ꆣ 춼쮵쏷쇋킭춬맽싋쯣램뗄풭샭ꆣ웤훐폃뮧1ꆢ2뫍3뚼뛔쿮쒿AꆢB뫍C뇭쿖돶탋좤ꎬ죧헢죽룶폃뮧뚼움볛쇋뗧펰AꆢB뫍Cꎬ헢훖훘뫏뇭쏷쯻쏇폐쿠쯆뗄탋좤ꆣ틲듋ꎬ뿉틔죏캪쿲폃뮧2췆볶D뫍E쫇뇈뷏뿉탐뗄ꎬ틲캪D뫍E뚼놻폃뮧1뫍3쯹쾲뮶ꆣ User3 GUser2 F ABC DE User1 춼ꎺ킭춬맽싋뗄릤ퟷ풭샭 킭춬맽싋쯣램뗄쫤죫폫쫤돶 폃뮧뛔룷훖짌욷뗄움볛쫽뻝ퟷ캪킭춬맽싋쯣램쓜릻ퟶ돶에뛏뗄틀뻝ꆣ샽죧듦퓚m 룶폃뮧{ u1ꎬu2ꎬꇵꎬum} 뫍n룶닺욷{ i1ꎬi2ꎬꇵꎬi n}ꎬ냑풭쪼폃뮧움럖쫽뻝뇭쪾돉캪m×n뗄뛾캬움럖뻘헳ꎬ벴폃뮧ꆪ닺욷쳵쒿뻘헳ꎨ죧뇭쯹쪾ꎩꆣ움럖뻘헳훐뗄횵Rꎺ쫇폃뮧i 뛔닺욷j쯹룸뗄움럖ꎬ춨뎣뇭쪾돉캪튻뚨i,j랶캧뗄헻쫽횵ꎬ샽죧1ꆫ5ꎻꎿ퓲듺뇭폃뮧뛔룃닺욷쎻폐ퟷ돶움럖ꎬ췆볶쾵춳폐틥컱뛔캴횪횵룸돶풤닢쳮돤ꆣ샻폃뛠훖춾뺶살웋ힽ폃뮧움럖쫽뻝ꎬ죧퓚GroupLens뫍Ringo쾵춳훐ꎬ폃뮧탨튪뛔쾵춳훐뗄짌욷쿔쪽움럖ꎬ떱좻ꎬ뷨훺폃뮧뗄붻틗볇슼ꆢ샀탐캪틾쪽췚뻲돶폃뮧움럖쫽뻝튲쫇뎣폃뗄춾뺶ꆣ킭춬맽싋췆볶쾵춳뗄쯣램뻍쫇캪캴움럖쿮쒿뗄풤닢움럖횵뗄맦퓲ꎬ맦퓲풽컇뫏쫂쪵ꎬ12
볆쯣뗄캴횪횵뻍풽헽좷ꎬ췆볶뗄킧맻뻍믡풽뫃ꆣ퓚뗧ퟓ짌컱췸햾훐훐ꎬ폃뮧움럖탅쾢뻘헳쫇튻룶룟캬쾡쫨뗄뻘헳ꎬ틲캪폃뮧횻믡움볛뫜짙튻늿럖탅쾢쳵쒿ꆣ룃컊쳢쫇킭춬맽싋쯣램탨튪쏦쇙훁맘훘튪뗄컊쳢ꆣ냑펦폃췆볶쾵춳쿫믱좡췆볶뷡맻뗄폃뮧돆캪쒿뇪뿍뮧ꎬ볇ퟷuꎬ췆볶쯣램뷡맻뗄탎쪽쫇ꎺ볆쯣돶쒿뇪뿍뮧u뛔쿮쒿i 뗄움럖횵ꎻ쇭튻훖쫇쒿뇪뿍뮧뿉쓜룐탋좤뗄N 룶닺욷뗄췆볶벯뫏ꎬ[24]떱좻헢N 룶닺욷벯뫏늻냼삨쒿뇪뿍뮧틑뺭릺싲뗄짌욷ꆣ 뇭 폃뮧ꆪ쿮쒿움럖뻘헳 쿮 쒿 돠뇚 짱뮰 뎤붭7뫅 컞벫 폃 뮧 샮쏷 ꎿ 3 2 6 췵컥4ꎿ33헅쟠 3 5 ꎿ 3 쇵뚬43ꎿ ꎿ 킭춬맽싋뗄폅뗣 [25]킭춬맽싋쯣램뇈웤쯼췆볶쯣램뗄폅쫆쳥쿖퓚틔쿂벸룶랽쏦ꎺ (1) 췆볶펦폃쇬폲맣ꎬ죧뛔쫩벮ꆢ틴샖ꆢ뗧펰뫍럾컱뗈췆볶ꆣ (2) 컼좡믹폚쓚죝맽싋뗄뷌통ꎬ폫웤쯼랽램돤럖죚뫏ꎬ뿉틔쪵쿖닺욷훊솿ꆢ럧룱뗈랽쏦뗄췆볶ꆣ (3) ퟔ쫊펦탔쓜솦잿ꎺ쾵춳췆볶훊솿뿉틔늻뛏쳡룟ꆣ (4) 뷢뻶탂폃뮧뫍탂쿮쒿뗄쓜솦잿ꆣꎨserendipitions recommendationsꎩ 킭춬맽싋쯣램뗄럖샠 떽쒿잰캪횹ꎬ퇐뺿죋풱쳡돶쇋킭춬맽싋뗄룷훖쪵쿖쯣램ꎬ틀뻝킭춬맽싋쯣램돶랢뗣뗄뇪ힼꎬ퓲뿉럖캪믹폚폃뮧- 폃뮧맘쾵뗄킭춬맽싋쯣램뫍믹폚쿮쒿- 쿮쒿맘쾵뗄킭춬맽싋쯣램ꆣ [26](1) 믹폚폃뮧- 폃뮧맘쾵뗄킭춬맽싋쯣램ꆣ 떱잰ꎬ퓚룃샠킭춬맽싋쯣램훐ꎬ펦폃ퟮ캪돉릦뗄췆볶벼쫵뻍쫇ꆰퟮ뷼쇚뻓쯣램ꆱꆣ헢훖쯣램샻폃춳볆랽램췚뻲돶폫쒿뇪폃뮧쿠쯆뛈ퟮ룟뗄죴룉폃뮧ꎬ돆캪ꆰ쇚뻓ꆱꎬ좻뫳틀뻝헢킩쇚뻓뗄움럖췆닢쒿뇪폃뮧뛔쒿뇪짌욷움럖횵ꆣ쒿잰ꎬ움볛폃뮧횮볤쿠쯆탔뗄뇪ힼ쫇ꎺ욤뛻즭ꎨPearsonꎩ쿠맘쾵쫽ꆢ폐풼쫸뗄욤뛻즭쿠맘쾵쫽ꆢ쿲솿쿠쯆탔뗈ꎬ틔벰붨솢헢킩ힼ퓲믹뒡짏뗄뇤쳥살뫢솿폃뮧횮볤뗄쿠쯆탔ꆣ웤훐ꎬ욤뛻즭쿠맘쾵쫽퓚킭춬맽싋쾵춳훐쪹폃뗃ퟮ캪맣랺ꎬ쫔퇩튲횤쏷쇋웤폐킧탔ꆣ짨폃뮧a 뫍폃뮧b릲춬움럖맽뗄쿮쒿벯뫏캪I=IΙIꎬ폃뮧a 뫍abab폃뮧b횮볤뗄Pearson쿠맘쾵쫽뗄볆쯣릫쪽죧쿂ꎺ 13
(R−R)(R−R)∑a,kab,kb k∈IabSim(a,b)=22R−RR−R∑a,ka∑b,kbk∈Ik∈Iabab웤훐ꎬR쫇뇭쪾폃뮧a뛔쿮쒿k뗄움럖ꎬR뫍R럖뇰쫇폃뮧a 뫍폃뮧b 쯹a,kab폐듲맽럖뗄쿮쒿뗄욽뻹뗃럖ꆣ 쇭튻훖랽램쫇붫맘폚폃뮧a뫍폃뮧b뗄볇슼뾴ퟷ쫇n캬쿮쒿뿕볤짏뗄솽룶쿲솿ꎬ폃뮧a 뫍폃뮧b횮볤뗄쿠쯆탔퓲뿉틔폃쿲솿횮볤볐뷇뗄폠쿒횵살뛈솿ꎬ폠쿒ρ횵풽듳뇭쏷솽룶폃뮧뗄쿠쯆돌뛈풽룟ꆣ폃뮧a 뫍폃뮧b뗄움럖쿲솿럖뇰뇭쪾캪aρ뫍bꎬ퓲폃뮧뗄쿠쯆탔Sim(a,b)캪ꎺ ρρρρa•b Sim(a,b)=cos(a,b)=ρρa×b웤훐ꎬ럖ퟓ캪솽룶폃뮧움럖쿲솿뗄쓚믽ꎬ럖쒸쫇쿲솿쒣뗄돋믽ꆣ 헒떽쇋뛈솿쿠쯆뛈뗄뇪ힼ횮뫳ꎬ뻍쓜췚뻲돶쒿뇪폃뮧l룶쇚뻓ꆣ뿉틔듓솽룶[29]랽쏦뾼싇ꎺꆣ ꋙ 틔쒿뇪폃뮧캪횧뗣ꎬ헒떽ퟮ쿠쯆뗄l룶폃뮧ꆣ ꋚ 닉폃뻛벯뗄랽램ꎬ쫗쿈헒돶뫍쒿뇪폃뮧ퟮ쿠쯆뗄폃뮧ꎬ좻뫳쮳탲헒돶1폠쿂뗄폃뮧ꆣ볙짨틑뺭헒떽j룶쇚뻓ꎬ볆쯣j룶쇚뻓뗄훐탄캻훃ꎬ톡퓱쪣c=V∑jj폠폃뮧훐폫룃훐탄캻훃ퟮ뷼뗄폃뮧ퟷ캪뗚j + 1룶쇚뻓ꆣ죧듋랴뢴ꎬ횱떽j =l 캪횹ꆣ헢훖랽쪽퓚쾡쫨쫽뻝벯훐뇭쿖뫜뫃ꆣ ퟮ뫳볆쯣룷쇚뻓뛔짌욷j움럖뗄볓좨뫍ꎬ틀뻝쪵볊돌뛈좷뚨좨횵듳킡ꎬ헢룶횵뻍뿉ퟷ캪쒿뇪뿍뮧u뛔쿮쒿i움럖뗄풤닢횵ꆣ뻟쳥뗄볆쯣릫쪽죧쿂ꎺ jP=R+ksim(u,i)(R−R) u∑,iuijii=1쪽훐ꎬk쫇폃살맦랶뮯좨횵뗄틲ퟓꆣ [24] (2) 믹폚쿮쒿- 쿮쒿맘쾵뗄킭춬맽싋쯣램ꆣ 룃킭춬맽싋쯣램뗄돶랢뗣쫇쿮쒿횮볤뗄쿠쯆뛈ꆣ쯣램쫗쿈헻샭돶쯹폐쒿뇪폃뮧틑뺭ퟷ돶움럖뗄쿮쒿벯뫏ꎬ볆쯣헢킩쿮쒿폫듽췆볶뗄쿮쒿i횮볤뗄쿠쯆돌뛈ꎬ늢듓훐즸톡돶쿠쯆뛈ퟮ룟뗄k룶쿮쒿{ i1ꎬ i2ꎬ ꇵꎬik}ꎬ웤뛔펦뗄쿠쯆돌뛈횵캪{ si1ꎬ si2ꎬ ꇵꎬsik}ꆣ좻뫳볆쯣쒿뇪폃뮧뛔헢킩쿠쯆쿮쒿움럖뗄볓좨욽뻹횵ꎬ벴뗃떽쯹탨뗄풤닢횵ꆣ쯣램뗄쪵쿖늽훨냼삨쿠쯆뛈볆쯣뫍볆쯣돶풤닢횵ꆣ 쿠쯆뛈볆쯣랽램뫍믹폚폃뮧횮볤맘쾵뗄ퟮ뷼쇚뻓쯣램샠쯆ꎬ쟸뇰퓚폚ퟮ뷼쇚뻓쯣램볆쯣폃뮧ꆪ쿮쒿움럖뻘헳뗄탐횮볤뗄쿠쯆뛈ꎬ뛸믹폚쿮쒿횮볤맘쾵뗄쯣램볆쯣쇐폫쇐횮볤뗄쿠쯆뛈ꆣ뻟쳥뗄맽돌쫇ꎺ튪볆쯣쿮쒿i폫쿮쒿j횮볤뗄쿠쯆뛈ꎬ뗚튻늽헒돶쯹폐춬쪱뛔헢솽룶쿮쒿움럖뗄폃뮧ꎬ믦훆믹폚솽룶쿮쒿뗄쇐쿲솿ꎬ뷓ퟅ톡퓱쿠맘쾵쫽램믲폠쿒횵램볆쯣솽룶쿲솿횮볤뗄쿠쯆뛈ꎬ쿠쯆뛈볆14
쯣릫쪽폫짏쫶샠쯆ꎬ늻퓙ힸ쫶ꆣ퓚폠쿒횵뗄랽램훐뮹튪뾼싇늻춬폃뮧움럖뇪ힼ뗄닮틬탔ꎬ쏖늹뗄냬램쫇듓쎿뛔횵훐복좥폃뮧뗄욽뻹횵ꎬ볆쯣릫쪽캪ꎺ (R−R)(R−R)∑k,ikk,jk k∈uSim(i,j)=22R−RR−R∑k,ik∑k,jkk∈uk∈u쪽훐u쫇쯹폐춬쪱뛔쿮쒿i뫍쿮쒿j움럖뗄폃뮧벯뫏ꆣ 럖뇰볆쯣돶쒿뇪폃뮧쯹폐틑뺭움볛뗄쿮쒿폫튪풤닢뗄쿮쒿i횮볤뗄쿠쯆뛈뫳ꎬ헒돶잰k룶횵ퟮ듳뗄쿮쒿Nꎬ좻뫳볆쯣헢킩횵뗄뫍ꎬ쎿룶횵틔쿠쯆뛈캪좨ꆣ 킭춬맽싋쯣램훐듽뷢뻶뗄컊쳢 [22ꎬ26]쪵퇩횤쏷킭춬맽싋쯣램췆볶뗄뿉뾿탔뷏룟ꎬ떫뮹폐컊쳢?탨뷢뻶:ꆣ ꎨ1ꎩ움럖쫽뻝쾡쫨탔컊쳢 뗧ퟓ짌컱췸햾훐짌욷쫽솿뻞듳ꎬ룶쳥폃뮧폐붻틗볇슼믲뛔짌욷ퟷ움볛뗄횻쫇웤훐벫짙뗄튻늿럖ꎬ룟캬쾡쫨움럖쫽뻝뻘헳듳듳붵뗍쯣램ힼ좷싊ꆣ췆볶쯣램훐볆쯣쿠맘쾵쫽훁짙탨튪솽룶폃뮧횮볤폐솽룶틔짏뗄훘뗾늿럖ꎬ틲캪쎻폐뛔쿠춬뗄짌욷움럖ꎬ쿠쯆뛈뫜룟뗄폃뮧뛸쪧횮붻뇛ꎬ췆볶쾵춳튲헒늻떽뿉릩췆볶뗄짌욷ꆣ 떱잰뎣폃뗄뷢뻶랽램쫇ꎺ샻폃캬맩풼벼쫵쿈뛔풭쪼움럖쫽뻝붵캬뒦샭ꎬ뒦샭뫳뗄쫽뻝ퟷ캪킭춬맽싋쯣램뗄쫤죫쫽뻝ꎬ떱좻붵캬뗄춬쪱뺡뿉쓜복짙탅쾢뗄쯰쪧ꆣ쒿잰뎣폃뗄캬맩풼벼쫵폐믹폚탅쾢퓶틦뗄쫴탔쿠맘럖컶ꆢSVD웦틬횵럖[27]뷢뗈ꆣ ꎨ2ꎩ샤뾪쪼컊쳢 샤뾪쪼컊쳢폖돆캪뗚튻움볛컊쳢(first rater) 믲탂쿮쒿컊쳢(new - item)ꎬ죧맻튻룶탂짌욷쎻폐죋좥움볛쯼ꎬ믲뚼늻좥움볛쯼ꎬ퓲헢룶쿮쒿뿏뚨뗃늻떽췆볶ꎬ췆볶쾵춳뻍쪧좥쇋ퟷ폃ꆣ헢쫇킭춬췆볶벼쫵폶떽뗄ퟮ캪춻돶컊쳢ꆣ쒿잰ꎬ뷢뻶뗄냬램훷튪쫇ퟩ뫏룷훖랽램ꎬퟮ캪뎣폃뗄쫇ퟩ뫏킭춬맽싋뫍믹폚쓚죝뗄췆볶ꆣ[34]Fab쾵춳뻍쫇믹폚헢튻쮼쿫붨솢뗄ꎬ쯼럖컶폃뮧움볛쿮쒿뗄쓚죝붨솢퓚믹폚쓚죝뗄폃뮧탋좤쇏ꎬ좻뫳춨맽폃뮧킭춬맽싋벼쫵랢쿖뻟폐쿠쯆탋좤낮뫃뗄쇚뻓폃뮧ꆣGroupLens퇐뺿킡ퟩ헽퓚펦폃폃뮧킭춬맽싋벼쫵좥ퟩ뫏웤쯻폃뮧뗄틢볻[35]뫍룶죋탅쾢맽싋듺샭ꆣ튲폐톧헟닉폃벤샸샭싛뗄랽램살뷢뻶샤뾪쪼컊쳢ꆣ ꎨ3ꎩ삩햹탔컊쳢 쯦ퟅ뗧ퟓ짌컱췸햾뗄엮늪랢햹ꎬ폃뮧탅쾢뫍닺욷쫽뻝돉놶퓶볓ꎬ췆볶쾵춳쯣램뗄킧싊늻뛏쿂붵ꎬ튪쟳늻뛏룄뷸쯣램쫊펦듳솿쫽뻝뗄볆쯣ꆣ뷢뻶뗄춾뺶쫇춨맽맽싋폃뮧붵뗍폃뮧쫽뻝솿ꆣ ꎨ4ꎩ쪵쪱탔 뗧ퟓ짌컱췸햾탨튪쪵쪱럾컱퓚쿟뿍뮧쫽솿뗄퓶볓쪹룃컊쳢뇤뗄풽살풽퇏뻾ꎬ뷨훺럖늼쪽볆쯣욽첨뿉틔뫜뫃뷢뻶룃컊쳢ꆣ쇭췢믹폚쿮쒿뻛샠뗄킭춬맽싋췆볶쯣램[28]뿉틔쳡룟닩헒ퟮ뷼쇚뻓뗄킧싊ꎬ냯훺췆볶쾵춳폐킧뷢뻶퓚뒦샭듳15
맦쒣쫽뻝쪱돶쿖뗄쪵쪱탔컊쳢ꆣ ꎨ5ꎩ좱랦풭쪼움럖쫽뻝 퓚뗧ퟓ짌컱췸햾퓋탐뗄맽돌훐ꎬ맽뛠뗘튪쟳폃뮧훷뚯쳡릩쫽뻝믡쪹폃뮧룐떽늻뇣뛸쿲웤쯻췸햾ꎬ튪믱뗃룼볓돤럖뗄폃뮧움럖쫽뻝ꎬ쒿잰닉폃뷏뛠뗄랽쪽쫇샻폃듺샭돌탲뛔폃뮧뗄샀탐캪ꆢ붻틗쫽뻝틾쪽볇슼ꎬ췚뻲폃뮧탋좤쒣탍ꎬ[25]Web쪹폃볇슼췚뻲벼쫵폐쇋폃커횮뗘ꎬ뮹뿉틔샻폃믹폚쿮쒿움럖풤닢뗄랽램쳮돤캴움럖쿮쒿ꆣ돽듋횮췢ꎬ쇩믮뗄틔뿍뮧캪훐탄췆볶탎쪽ꎺ췆볶쾵춳춨맽폃[22]뮧뗄뻛샠쪵쿖쾵춳췆볶ꎻ죃폃뮧닎폫떽췆볶뗄맽돌훐ꎬ췆볶뗄쮵럾솦룼잿ꆣ ꎨ6ꎩ붡ힳ탔럖컶 폐쪱짌볒캪쇋뺺헹뗄탨튪ꎬ웆뮵뺺헹뛔쫖뗄췆볶쾵춳ꎬ죋캪뗘훆퓬볙쫽뻝ꆣMichael O’Mahony 뗈싊쿈ퟶ쇋뾪뒴탔뗄릤ퟷꎬ컄쿗[35]폃룅싊랽램듓ힼ좷탔뫍컈뚨탔솽랽쏦뛔췆볶쾵춳붡ힳ탔뷸탐쇋럖컶ꎬ룸돶쇋솽훖럖컶쒣탍ꎬ뗃돶쇋퓚뾼싇볙쫽뻝뗄쟩뿶쿂췆볶헽좷뗄룅싊ꎬ떫듋럖컶폐뫜뛠뗄볙짨쳵볾ꆣ 떱잰뷢뻶컊쳢뗄랽램 킭춬맽싋뗄ퟔ짭쳘뗣틽웰짏쫶튻쾵쇐컊쳢ꎬힼ좷쿪쪵뗄폃뮧탅쾢쫇믱뗃샭쿫뗄췆볶킧맻뇘놸쳵볾ꆣ떫쒿잰뛸퇔ꎬ믱좡듳솿뗄폃뮧탅쾢쫽뻝뗄쓑뛈벫듳ꎬ떼훂킭춬맽싋벼쫵펦폃쇬폲뫜쿁햭ꎬ쟒벯훐퓚폩샖랽쏦ꎬ죧쫩벮ꆢ틴샖ꆭꆭꆣ죧퓚컄놾쿠맘탔쇬폲ꎬ킭뗷맽싋뫍믹폚쓚죝맽싋뇈뷏뇭쿖뗄닮잿죋틢ꆣ뷢뻶짏[30]쫶컊쳢뗄랽램폐솽쳵ꎺ ꋙ 퇐뺿탅쾢췚뻲벼쫵ꎬퟮ듳돌뛈뗘듓뷏짙뗄풭쪼움럖쫽뻝훐췚뻲폐폃뗄탅쾢ꆣ ꋚ 퇐뺿탅쾢믱좡벼쫵ꎬ헒떽룼뛠뗄쳡좡뫏샭폐킧폃뮧탅쾢뗄춾뺶ꆣ 탅쾢췚뻲벼쫵뗄퇐뺿 쒿잰킭춬맽싋쾵춳뗄돵쪼탅쾢살ퟔ폚폃뮧뛔쿮쒿뗄쿔쪽움럖쫽뻝ꎬ춨뎣뷨훺뛾캬뗄폃뮧튻쿮쒿움럖뻘헳뇭쪾ꆣ죧뫎틔룃움럖뻘헳캪욽첨ꎬ췚뻲돶폃뮧룼짮닣듎뗄룶탔뮯쳘헷쫇탅쾢췚뻲벼쫵랽쿲퇐뺿뗄룹놾쒿뇪ꆣ 늻춬폃뮧움볛쿮쒿뗄훘뗾늿럖쫇볆쯣폃뮧쿠쯆뛈뗄잰쳡ꎬ죧맻훘뗾뗄움볛쿮쒿첫짙ꎬ뗃떽풭쪼뗄움럖탅쾢컞램뻶뚨폃뮧횮볤뗄쿠쯆탔ꆣ쫂쪵짏ꎬ폃뮧튻쿮쒿움럖뻘헳쫇튻룶룟캬쾡쫨뻘헳ꎬ횱뷓펰쿬쿠쯆탔쯣램뗄뿉뾿탔ꆣ쳘뇰퓚튻킩벫뛋쟩뿶쿂ꎬ죧뛔튻룶췆볶쾵춳뗄탂폃뮧ꎬ쎻폐웤쯻폃뮧폫웤릲춬움럖뗄쿮쒿ꎬ쾵춳뻍컞램캪탂폃뮧쳡릩췆볶ꆣ 컒쏇뿉틔톰헒탂뗄춾뺶ꎬ죧폃뮧A뫍폃뮧B뗄쿠쯆탔잿ꎬB폖뫍C튲폐뷏룟뗄쿠쯆뛈ꎬ헢뇭쏷A뫍C횮볤듦퓚튻뚨뗄쿠쯆탔ꎬ떫A뫍C죧맻쎻폐훘뗾뗄움럖쿮쒿ꎬ뒫춳뗄킭춬맽싋쯣램뻍랢쿖늻쇋A뫍C뗄맘쾵ꆣ늻뛏췪짆헢킩탅쾢폐킧뗘샻폃랽램ꎬ삩듳킭춬맽싋벼쫵펦폃쇬폲ꎬ룼맘볼퓚폚쳡룟췆볶쾵춳풤닢ힼ좷뛈ꆣ 듓쿮쒿쳵쒿뗄뷇뛈뾴ꎬ움럖뻘헳훐ꎬ쿮쒿뗄쿠쯆탔쳥쿖퓚탐쿲솿쿠맘(폃뮧16
뛔쿠쯆쿮쒿뗄움럖횵뷓뷼)ꆣ뷨훺듋쮼쿫톹쯵움럖뻘헳늢뛔웤붵캬뒦샭ꎬ돤럖췚뻲퓌몬탅쾢뗄쓚퓚솪쾵ꆣ듋랽램폫뒫춳뗄IR(Information Retrieval)훐뗄틾몬폯[31]틥쯷틽(LSI)폐뫜듳뗄쿠쯆탔ꆣ퓚킭춬맽싋벼쫵훐ꎬ샻폃웦틬횵럖뷢(SVD)릤뻟ꎬ붵뗍움럖뻘헳캬쫽ꎬ돤럖랢뻲풭쪼쫽뻝탅쾢쳵쒿횮볤잱퓚솪쾵ꆣ톧헟뗄퇐[32]뺿훐퇩횤쇋짏쫶쮼슷뗄킧맻ꆣ 탅쾢믱좡벼쫵뗄퇐뺿 퓚벺폐뗄풭쪼움럖쫽뻝훐ꎬ탨튪돤럖췚뻲쫽뻝쓚늿퓌몬뗄맦싉ꎬ돽듋횮췢ꎬ뮹튪톰헒룼뛠룼폐킧뗘웋ힽ폃뮧탅쾢뗄춾뺶ꆣ 폃뮧뗄탋좤쒣탍늻뷶뿉틔뷨훺폃뮧쿔쪽움볛뗃떽ꎬ뮹뿉틔듓탎쪽뛠퇹뗄틾쪽움볛훐쳡좡ꆣ떱좻듓폃뮧틾쪽움볛쒣쪽훐췪돉폃뮧탋좤쒣탍뗄붨쒣ꎬ뮹쏦쇙ퟅ뫜뛠뗄컊쳢ꎬ죧ꎺ틾쪽움볛뗄샠탍ꎿ틾쪽움볛뗄뿉뾿탔ꎿ틾쪽움볛탅쾢뫍쿔쪽움볛탅쾢폐킧뷡뫏뗄랽쪽뗈뗈ꆣ헫뛔틾쪽움볛뗄퇐뺿릤ퟷ뷶뷶삭뾪탲쒻ꎬ뫜뛠뗄컊쳢뮹쎻폐탐횮폐킧뗄뷢뻶랽램ꆣ뮹뿉틔짨쿫ꎬ폐쎻폐웤쯼탅쾢뿉틔냯훺췆볶쾵춳볆쯣폃뮧횮볤뗄쿠쯆탔ꆣ샽죧솽폃뮧뗄쓪쇤쿠럂ꎬ쯻쏇뿉쓜뛔쒳킩탅쾢폐릲춬탋좤ꆣ헢킩폃뮧뗄럇움럖탅쾢뛔에뛏쿠쯆폃뮧폐뫜뫃뗄킧맻ꆣShapira[33]뫍Shoral짨볆뗄탅쾢맽싋쾵춳뛔룃쮼슷뷸탐폐틦뗄뎢쫔ꎬ떱좻ꎬ뮹탨뷸탐뷸튻늽뗄퇐뺿ꆣ 킭춬맽싋벼쒾냑뿉뾿탔탅쾢췆볶ퟷ캪웤훕벫쒿뇪ꎬ횻폐폐믺죚뫏탅쾢맽싋랽램ꆢ탅쾢췆볶벼쫵ꎬ닅폐룼듳뗄랢햹뿕볤ꆣ믹폚듋탭뛠탂뗄벼쫵폫랽램펦퓋뛸짺ꎺ럖샠랽램ꆢ춼싛랽램ꆢweb췚뻲뫍믹폚쿲솿믺ꎨSupport vector machineꎮꎩꎬ뿉틔뷨훺뇭룱살뾴튻쿂뻟쳥뗄벼쫵ꎺ 뇭 뻟쳥킭춬맽싋벼쫵뗄훖샠 Collaborative Techniques Typical papersKNN algorithm Resnickꎬ1994ꎻ and improved one Mohanꎬ et al 2006 Choaꎬ et alꎬ 2004ꎻ Web mining Choꎬ et alꎬ 2002 Decision tree Choꎬ et alꎬ 2002 Graph theory Aggarwalꎬ 1999 Heuristic-based method Support vector machine Min & Hanꎬ 2005 Breeseꎬ et alꎬ 1998ꎻ Bayesian model Chienꎬ et alꎬ 1999 Breeseꎬ et alꎬ 1998ꎻ Clustering Goldbergꎬ et alꎬ 2001 Association rule mining Kimꎬ et alꎬ 2004 Artificial neural Pazzaniꎬ et alꎬ 1997 network Linear-regression Vuceticꎬ et alꎬ 2005 Maximum Model-based Pavlovꎬ et alꎬ 2002 entropy method Latent semantic Hofmannꎬ 2004ꎻ analysis Markov process Shaniꎬ et alꎬCheungꎬ et alꎬ 2004 2002 17
놾헂킡뷡 뗧ퟓ짌컱췸햾훐ꎬ췆볶쾵춳볈뿉틔냯훺폃뮧퓚훚뛠뗄짌욷훐뿬쯙뚨캻룐탋좤뗄짌욷ꎬ튲뿉틔캪뗧ퟓ짌컱쳡릩짌쪵쿖샻죳ꎬ틑랢햹돉캪뗧ퟓ짌컱췸햾뗄늻뿉좱짙뗄ퟩ볾ꆣ킭춬맽싋쯣램폐웤쯻췆볶랽램컞램뇈쓢뗄폅뗣ꎬ놾헂쫗쿈볲떥뷩짜쇋킭춬맽싋쯣램뗄믹놾횪쪶ꎬ좻뫳쿪쾸뷩짜쇋킭춬맽싋췆볶쯣램뗄샠탍ꎬ냼삨믹폚폃뮧뗄킭춬맽싋쯣램뫍믹폚폃뮧뗄킭춬맽싋쯣램ꎻ늢뛔듦퓚뗄컊쳢뷸탐쇋룅쫶ꎬ훷튪뇭쿖퓚ꎺ쫽뻝쾡쫨탔컊쳢ꎻ샤웴뚯컊쳢ꎻ삩햹탔컊쳢ꎻ쪵쪱탔ꎻ움럖쫽뻝늻ퟣ뫍쾵춳뗄붡ힳ탔컊쳢ꆣퟮ뫳럖컶쇋떱잰뷢뻶컊쳢뗄랽램폫쮼슷ꆣ컒맺뗧ퟓ짌컱췸햾뗄췆볶쾵춳죔뒦퓚뿬쯙랢햹뗄뷗뛎ꎬ퇐뺿룼뫃킭춬맽싋쒣탍벰쯣램죎훘뛸뗀풶ꆣ18
뗚쯄헂 믹폚쮫탅쾢풴뗄킭춬맽싋쯣램퇐뺿 쿠맘믹뒡횪쪶 믹폚쮫탅쾢풴뗄킭춬맽싋쯣램뗄놳뺰 [37]퓚뗚죽헂컒쏇럖컶쇋뒫춳CF벼쫵폶떽튻킩벬쫖뗄컊쳢ꆪ샤웴뚯컊쳢ꆢ쫽뻝뗄쾡쫨탔컊쳢ꆢ췆볶뗄뿉뾿탔컊쳢뗈뗈ꆣ쳘뇰쫇ꎬCF퓚뒦샭붻뮥탔잿ꆢ탨튪쏅벼쓜횪쪶뗄쇬폲ꎬ룼쿔뗃솦늻듓탄ꆣ 캪뷢뻶짏쫶컊쳢ꎬ쳡룟CF벼쫵뗄킧쓜ꎬ퇐뺿헟쳡돶쇋튻킩룄뷸뒫춳CF뗄쯣램ꆣ컄쿗[38]쳡돶믹폚쒣쪽ꎨModel-basedꎩCF 쯣램ꎬ샻폃웓쯘뗄BayesꎨNBꎩꆢtree—argument NBꆢNB-ELR믲헟TAN-ELR럖샠웷퓚뒦샭늻췪좫쫽뻝랽쏦뗄쓜솦ꎬ뛔폃뮧ꆪ쿮쒿뻘헳훐좱쪧뗄움럖쪵쿖쳮돤ꎬ헢훖랽램퓚뷢뻶쫽뻝뗄쾡쫨탔컊쳢폐튻뚨뗄킧맻ꎬ떫쫇ꎬ쿖폐뗄퇐뺿뷡맻뇭쏷ꎬ뛔뒫춳CF쯣램탔쓜뗄룄뇤뗄틢틥늻듳ꆣ컄쿗[39]쳡돶믬뫏뗄췆볶쾵춳ꎨhybrid CF algorithmꎬHCFꎩꎬ쫔춼튪뿋럾떥룶쾵춳뗄좱뗣ꆣퟮ폐듺뇭탔HCF쫇content-boosted CFꎬ펦폃NB췸싧ꎬ쳮돤CF움럖뻘헳훐좱쪧뗄쫽뻝쿮ꎬ릹돉튻룶늻췪좫헦쪵뗄폃뮧움럖뻘헳ꆣ좻뫳ꎬ뷨훺듸폐좨훘뗄Pearson correlation펦폃떽헢룶폃뮧움럖뻘헳훐ꎬ탎돉쳘쫢뗄폃뮧ꆪ쿮쒿뗄췆볶ꆣ쫔퇩퇩횤쇋헢룶쯣램뗄폅풽탔ꎬ떫룃쯣램쎻폐뾼싇탨튪쏅횪쪶쇬폲뗄췆볶쿮쒿ꎬ뛸쟒쿮쒿뗄쓚죝탅쾢늢늻쫇뚼쓜뗃떽ꎬ쳘뇰쫇뛔폚폐탭뛠뗄닺욷훖샠듳탍뗧ퟓ짌컱췸햾뛸퇔ꎬ캪쇋쪵쿖content-boosted CFꎬ탨튪쳡좡헢킩닺욷훖샠뗄탭뛠쳘헷ꎬ헢쿮릤ퟷ쫇럇뎣삧쓑뗄ꆣ 떽쒿잰캪횹ꎬ쯹폐CF쯣램냑쯼쏇뗄췆볶붨솢퓚튻룶떥튻뗄췆볶ퟩ훐ꎬ캪쇋뷢뻶쫽뻝쾡쫨탔ꆢ헫뛔뻟폐쏅횪쪶놳뺰뗄쿮쒿췆볶ꎬ놾컄쳡돶믹폚쮫탅쾢풴쒣쪽뗄킭춬맽싋쯣램ꎨDual Information Source Model-Based Collaborative Filtering AlgorithmsꎬDISCFꎩꎬ룃랽램에뛏믮뚯폃뮧뛔쒿뇪쿮쒿뗄탋좤돌뛈붨솢퓚솽룶췆볶ퟩꆪꆪ쿠쯆폃뮧췆볶ퟩꎨퟮ뷼쇚뻓벯뫏ꎩ폫볒췆볶ퟩ믹뒡짏ꎬ냑솽룶췆볶ퟩ뗄붨틩뷡뫏웰살ꎬ탎돉뿉뾿뗄탅쾢풴ꎬ좻뫳ꎬ럖컶룷ퟔ펰쿬믮뚯폃뮧뛔쒿뇪쿮쒿뗄좨훘ꎬ볆쯣믮뚯폃뮧뗄ퟮ훕탋좤뛈ꎬ쪵쿖쾵춳췆볶ꆣ DISCF쯣램뗄쿻럑탄샯톧믹뒡 DISCF쯣램쫇붨솢퓚쿻럑헟탄샯톧맛뗣믹뒡짏뗄ꎬ펰쿬쿻럑헟릺싲탐캪뗄틲쯘뫜뛠ꎬ헢킩틲쯘닺짺뗄ퟷ폃튲늻쿠춬ꆣ샽죧ꎺ닺욷뗄쳘탔ꆢ닺욷폫쿻럑헟[40]뗄쏜쟐맘쾵뫍쿻럑헟뛔닺욷뗄쇋뷢ꆣ 퓚쿖쪵짺믮훐ꎬ튻룶쿻럑헟퓚톡퓱뗧펰믲릺싲죕폃욷뗈킡탍짌욷쪱뿉쓜뷶뷶헷쟳뻟폐쿠춬탋좤낮뫃뗄쿠쯆폃뮧ꎨퟮ뷼쇚뻓벯ꎩ뗄틢볻ꎬ떫쫇ꎬ떱쯻쏇퓚톡퓱볛룱뇈뷏낺맳뗄닺욷쪱ꎬ샽죧ꎺ튻첨뇊볇놾뗧쓔ꎬ췹췹룼뛠헷톯튵죋쪿뗄틢볻ꎬ헢뻍쫇췆볶탅쾢풴뗄뛠퇹탔ꎻ늻춬뗄쿻럑헟퓚쿠춬닺욷쇬폲훐뛔췆볶풴뗄톡퓱튲쫇늻춬뗄ꎬ폐뗄쿻럑헟믡톡퓱폐ퟅ쿠춬탋좤뗄ퟮ뷼쇚뻓뗄맛뗣ꎬ폐19
뗄믡룼뛠듓튵죋쪿짭짏믱좡붨틩ꎬ헢뻍쫇췆볶탅쾢풴뗄뿉탅뛈ꎻ폐뗄닺욷쿻럑헟풸틢뮨듳솿뗄쪱볤폫뺫솦좥쇋뷢닺욷뗄쳘탔ꎬ폐킩쿻럑헟쮲볤뻍쓜ퟶ돶릺싲뻶뚨ꎬ헢듺뇭닺욷뗄맘솪뛈쮮욽ꆣ쿖폐뗄CF쯣램쎻폐ퟛ뫏뾼싇췆볶풴뗄뛠퇹탔ꆢ뛠퇹뗄췆볶풴뗄탅죎뛈폫닺욷뗄훘튪탔쮮욽ꆣ DISCF쯣램쫽뻝ힼ놸 쪵쿖DISCF쯣램뗄맘볼쫇헒떽믮뚯폃뮧뗄쿠쯆폃뮧벯뫍볒폃뮧벯ꎬ늢럖뇰쟳돶뛔쒿뇪쿮쒿뗄췆볶횵틔벰췆볶횵뗄뿉탅뛈ꆣ퓚헢샯탨튪샻폃폃뮧ꆪ쿮쒿움럖뻘헳ꎬ떫뛔폚튻룶듳탍뗧ퟓ짌컱췸햾ꎬ폃뮧ꆪ쿮쒿움럖뻘헳쫇럇뎣쾡쫨ꎬ캪쳡룟볆쯣쿠쯆폃뮧뗄ힼ좷싊ꎬ뷢뻶쫽뻝쾡쫨탔ꎬ놾컄폃TAN-ELR랽램뛔폃뮧ꆪ쿮쒿움럖뻘헳훐캴움럖쿮ퟶ돶풤닢ꆣ 쿖폐뗄믬뫏뗄췆볶쾵춳ꎨhybrid CF algorithmꎬHCFꎩ뛠쫽폃놴튶쮹췸싧ꎨNBꎩퟷ캪쓚죝뗄풤닢웷ꆣNB뗄컊쳢퓚폚룸뚨뗄럖샠닎쫽훐ꎬ냑쫴탔ퟷ캪뛀솢뗄쓚죝살뾼싇ꎬ뛸a tree augmented naive Bayes network (TAN) 쓜릻돤럖뾼싇쫴탔횮볤뗄맘솪ꎬ헢퇹ꎬ뻍쓜뇈NB닺짺룼뫃뗄럖샠킧맻ꆣ좻뛸ꎬNB뫍TAN뚼탨튪톰헒닺짺ퟮ뫃킧맻뗄럖샠웷ꎬ뛸컒쏇ퟮ뫃뗄쒿뇪쫇쫂컯뗄본뇰웷ꆪLogistic regression(LR)ꎬ쯼쓜헒떽쪹본뇰킧맻듯떽ퟮ뫃뗄닎쫽ꎬ떫쫇LR탨튪튻룶쿱NB췸싧튻퇹볲떥뗄뷡릹ꆣTAN-ELR냑쯼쏇횮볤뗄폅뗣뷡뫏쇋웰살ꎬ쫗쿈헒떽튻룶쓜쪹본뇰킧맻ퟮ폅뮯뗄NB뷡릹ꎬ좻뫳헒떽튻룶킧싊룟뗄럖샠닎쫽ꆣGreiner et [38]al횤쪵쇋TAN-ELR퓚쏦뛔췪좫믲늻췪좫뗄쫽뻝쪱뚼뇭쿖돶뫜뫃뗄럖샠킧맻ꆣ틲듋ꎬ놾컄훐컒쏇쪹폃TAN-ELRퟷ캪쓚죝뗄풤닢웷ꆣ폃뮧풭폐움럖뇭ퟷ캪뗈벶횵ꎬ냑맘솪폃뮧뗄탅쾢ퟷ캪쫴탔횵살통솷TAN-ELR쒣쪽ꎬ춨맽헢훖쒣쪽뛔캴움럖뗄쿮쒿ꎬ뷡뫏쯼쏇뗄쓚죝탅쾢룸돶움럖뗄풤닢횵ꆣ퓚움럖뻘헳훐뛔쎿튻쇐ꎨ쎿튻룶쿮쒿ꎩ뚼훘뢴짏쫶맽돌ꎬ퓚움럖뻘헳훐캴놻움럖뗄쿮쒿믡놻폃헢훖쒣쪽뗃떽뗄풤닢횵쳮싺듓뛸뗃떽튻룶룄짆뗄폃뮧움럖뻘헳ꎬ뷢뻶쇋쫽뻝쾡쫨탔뗄컊쳢ꆣ DISCF쯣램쇷돌춼 DISCF쯣램쇷돌춼볻춼ꆣ 20
Contented-based predictor 쿠쯆폃 Person CF 쯣램 폃뮧쒣탍쫽(TAN-ELR) 쳮돤뫳뗄폃뮧벯뫏 폃뮧쪹폃 뻝뿢+폃뮧쒣탍쫽쫽뻝뿢 뮧탅쾢 뻝뿢 듸폐뿉탅뛈쿠 쯆폃뮧뗄췆볶 폃뮧쒣탍 닺짺쮫닺짺쎿룶쫽쫽뻝뿢 쫽뻝풴뻝풴뗄뿉탅췆볶쒿뇪 뛈쫽뻝뿢 듸폐뿉탅뛈 볒폃볒뗄췆뮧벯뫏 볶 춼ꎺDISCF쯣램쇷돌춼 DISCF쯣램뗄쪵쿖 릹퓬DISCF쯣램뗄쮫탅쾢풴 뺭맽뷚쫽뻝ힼ놸ꎬ쫗쿈ꎬ춨맽볆쯣쒿뇪폃뮧a뗄ퟮ뷼쇚뻓벯ꎬ늢쟒쟳돶쯼쏇뛔쒿뇪쿮쒿i뗄탋좤뛈ꎬ살릹퓬쿠쯆폃뮧뗄탅쾢풴ꆣퟮ뷼쇚뻓벯쫇횸폫쒿뇪폃뮧a폐쿠춬뗄폃뮧탋좤움럖뇭ꆣ퓚쿠쯆뛈볆쯣짏ꎬ닉폃퓚뗚죽헂쏨쫶뗄ꆢ퓚뒫춳뗄CF쾵춳훐펦폃ퟮ맣랺뗄Pearson’s맘솪쾵쫽ꎬ쒿뇪폃뮧a폫웤쯻폃뮧u뗄쿠쯆뛈뚨틥캪Wꎬ죧릫쪽쯹쪾ꎺ (u,a)sm(R−R)(R−R)∑u,iua,iai=1w(u,a)= s22mm(R−R)(R−R)∑u,iu∑u,iui=1i=1R헢샯쫇폃뮧u뛔쿮쒿i뗄움럖ꎬR쫇쒿뇪폃뮧뛔쿮쒿i뗄움럖ꆣ퓚컄쿗[41]u,ia,i뗄퇐뺿믹뒡짏ꎬ톡퓱쿠쯆뛈쫽횵뎬맽뗄폃뮧탎돉쒿뇪폃뮧뗄ퟮ뷼쇚뻓벯ꆣ쟳돶ퟮ뷼쇚뻓벯뛔쒿뇪쿮쒿i뗄풤닢횵S(aꎬi)ꎬ헢샯샻폃쿠쯆폃뮧뗄쳘탔볓좨쫽횵ꎬ뛸늻쫇볲떥뗄쿠쯆폃뮧탋좤뛈뗄욽뻹횵ꎬ웤탎쪽죧릫쪽쯹쪾ꎺ n(R−R)w(u,a)∑u,iusu=1 s(a,i)=(u,a)∑su=1 헢샯R쫇쇚뻓폃뮧u뛔쿮쒿i뗄움럖횵ꎬn듺뇭쒿뇪폃뮧a쇚뻓폃뮧뗄룶쫽ꆣ u,i좻뫳ꎬ헒떽볒폃뮧벯뫏ꎬ볆쯣돶쯻쏇뗄풤닢횵ꎬ탎돉뗚뛾룶탅쾢풴ꆪ볒탅쾢풴ꆣ퓚컄쿗[42]뫍샭뷢튵횪쪶룅쓮뗄믹뒡짏ꎬ놾컄룸돶볒폃뮧ퟩ(expert-users)뗄뚨틥ꎺ퓚냼몬듽췆볶쿮쒿뗄쇬폲훐ꎬ틑뺭ퟶ쇋듳솿뗄릤ퟷꎬ쯻쏇폐ퟣ릻뗄쓜솦뛔웤쯻폃뮧ퟶ돶헽좷뗄췆볶붨틩ꎬ늢짨볆돶튻룶쓜랴펳폃뮧쓜솦믲풤닢쓜솦뗄튵횪쪶벼쓜뗄뛈솿랽램ꆣ튵횪쪶벼쓜뿉틔쏨쫶캪헻쳥쮮욽21
ꎨtotal-item levelꆢ뗧펰볒ꎩꆢ샠뇰쮮욽ꎨcategory levelꆪ뚯ퟷ뗧펰볒ꎩ뫍룶뇰쮮욽ꎨindividual-item levelꎬ뗧펰돠뇚뗄볒ꎩꆣ컄쿗[43]훐쮵쏷퓚쪵쿖췆볶뗄맽돌훐ꎬ풽뻟쳥쏨쫶뗄볒ꎬ풽쓜쪵쿖룼뫃뗄췆볶ꆣ좻뛸ꎬ퓚쿖쪵짺믮훐ꎬ럇뎣뻟쳥ꆢ룶뇰뮯쮮욽뗄튵횪쪶뛔킭춬췆볶틢틥늻듳ꆣ ω틲듋ꎬ놾컄훐뛈솿튵횪쪶벼쓜쫇폃샠뇰쿮쒿쮮욽ꎨcategory levelꎩꆣ폃e살뚨틥볒횪쪶쮮욽ꎬ쯼뇭쪾폃뮧u뛔샠뇰c뗄튵횪쪶벼쓜쮮욽ꎬ쏨쫶죧릫쪽ꎺ R−R∑∑u,ja,jj∈c(i)a∈u(j)ω(u,c)=β(u,c)1− eNc(i)퓚헢샯U(j)쫇돽쇋폃뮧u횮췢뗄ꎬ뛔쿮쒿j폐닙ퟷ탐캪뗄쯹폐폃뮧뗄벯뫏ꎬC(i)쫇폫쒿뇪쿮쒿i쫴폚춬튻훖샠뗄쯹폐web훐놻럃컊맽쯹폐쿮쒿뗄벯뫏ꎬNc(i)쫇C(i)뗄믹쫽ꎬβ(u,c)쫇튻룶뿉틔뇤뮯뗄좨훘횵ꆣβ(u,c)뚨틥캪1 – 1/nꎬꎨn쫇퓚샠뇰훐놻움럖뗄듎쫽ꎩꆣ떱튻룶폃뮧퓚샠뇰훐움럖뗄듎쫽풽뛠ꎬ컒쏇뻍쓜뗃떽튻룶뷏룟뗄횪쪶쮮욽뗄횵ꆣ듓쎿튻룶샠뇰쇬폲훐뻟폐튵벼쓜횪쪶뗄폃뮧훐톡퓱ퟮ룟뗄3%살ퟩ돉볒폃뮧ퟩꆣ볒폃뮧ퟩ뗄췆볶쓜솦폃E(i)살뇭쪾ꎬ웤탎쪽쏨쫶죧릫쪽ꎺ n(R−R)w(u,c)∑u,iueu=1E(i)= nw(u,c)∑eu=1헢샯ꎬR쫇볒폃뮧u뛔쒿뇪쿮쒿i뗄움럖ꎬn쫇퓚볒폃뮧ퟩ훐볒뗄쫽u,i쒿ꆣ DISCF쯣램쮫탅쾢풴뗄뿉탅뛈볆쯣폫췆볶뗄탎돉 뺭맽뗄볆쯣ꎬ믱뗃쎿룶탅쾢풴뛔쒿뇪폃뮧뫍쒿뇪쿮쒿췆볶횵ꎬ쿂튻늽볆쯣쒿뇪폃뮧뛔솽룶탅쾢풴뿉탅뛈ꆣ볙짨쎿튻룶폃뮧뛔죎틢뗄탅쾢풴뗄뿉탅뛈뚼늻튻퇹ꎬ놾컄쳡돶쿂쏦뗄췆볶풴뿉탅뛈쒣쪽죧릫쪽ꎺ R(a,i)−R=KS(a,i)+KE(i)+ past−ratingasea,ia웤훐ꎬ듺뇭쒿뇪폃뮧폫쿮쒿뗄뇠뫅ꎬK폫K럖뇰듺뇭쒿뇪폃뮧뛔seퟮ뷼쇚뻓벯폫볒폃뮧벯뿉탅뛈ꆣ컒쏇뿉틔샻폃ퟮ킡뛾돋램뗄뛠훘믘맩믽럖럖컶뗄랽램쟳돶쯼쏇뗄횵ꆣ죧맻쿠쯆폃뮧폫볒폃뮧폐훘뗾뗄늿럖ꎬS(a,i)뫍E(i)뻍믡랢짺뛠훘릲쿟탔컊쳢ꎬ퓚헢훖쟩뿶쿂ꎬ볆쯣K폫뻍뿉쓜뇤뗄뫜삧쓑ꆣKse죧맻퓚솽룶뇤솿횮볤듦퓚뛠훘릲쿟탔ꎨ떱엲헍틲ퟓ횵듳폚10쪱ꎬ돥춻듦퓚ꎩꎬ컒쏇뻍쪹폃믘맩럖컶살볆쯣K뫍Kꆣퟮ뫳췪돉뛔췆볶쿮쒿뗄풤닢움럖ꎬ뇭쪾탎se쪽죧릫쪽ꎺ P(a,i)=R+KS(a,i)+KE(i)+ predictionase퓚쪵쿖췆볶뗄맽돌훐ꎬ뿉쓜믡폶떽솽훖쳘쫢뗄쟩뿶ꎺꎨ1ꎩ횻듦퓚튻룶탅쾢22
풴ꎨ샽죧ꎺ쎻폐ퟮ뷼쇚뻓벯ꎩꎬDISCF쯣램붫퓚듦퓚뗄탅쾢풴훐좥쟳뛔펦뗄뿉탅뛈ꆣꎨ3ꎩ탂폃뮧믲헟탂쿮쒿뗄췆볶ꎬ캪튻룶탂폃뮧헒떽ퟮ뷼쇚뻓쫇럇뎣삧쓑뗄ꎬDISCF쯣램뛔탂폃뮧뗄췆볶횻쓜뷨훺볒폃뮧ퟩ뗄췆볶붨틩ꆣ춬퇹ꎬ퓚튻룶쿮쒿뗄돵쪼ힴ첬쿂ꎬ헒떽폫횮맘솪뗄폃뮧튲뇤뗄뫜삧쓑ꎬ뛔탂쿮쒿뗄췆볶ꎬ튲탨튪쟳훺볒탅쾢풴뗄췆볶붨틩ꆣ DISCF쯣램쪵쿖맽돌 쫤죫ꎺRꎺ폃뮧탋좤쒣탍 Kꎺ탅쾢풴뗄뿉탅죎뛈 Pa,i) 쫤돶ꎺ( 췆볶뗄붨틩 main() {for 쯹폐뗄쿮쒿i for 쯹폐뗄폃뮧u ifꎨu==newuserꎩthen call NewUserRecom(); else if(i==newitem) then call NewItemRecom(); else call ExUserRecom(); endif ennfor endfor} ExUserRecom() Begin 솽룶췆볶ퟩ뗄풤닢횵E(i)뫍S(a,i) cEi)Kifꎨ(뫍S(a,i)뚼듦퓚ꎩ톡퓱Kꆢ뫍 se볆쯣돶ꎺ P(a,i)=R+KS(a,i)+KE(i)+c aseElse cE(i) ifꎨ늻듦퓚ꎩthen 톡퓱K폫 ss0 //K쫇뷶폐튻룶쿠쯆폃뮧탅쾢풴뗄뿉탅뛈 s0 P(a,i)=R+KS(a,i)+c as0sc ifꎨS(a,i)늻듦퓚ꎩKthen 톡퓱폫 eie0 //K쫇뷶폐튻룶볒폃뮧탅쾢풴뗄뿉탅뛈 e0P(a,i)=R+KE(i)+c ae0eiEnd if End ExUserRecom() 23
NewUserRecom() Begin E(i)볆쯣볒ퟩ뗄풤닢횵 볆쯣돶 P(a)RE( ,i=+i)uR //쯹폐폃뮧움럖뗄욽뻹횵 uEnd NewUserRecom() NewItemRecom() Begin 볆쯣볒ퟩ뗄풤닢횵E(i)ꎬ 톡퓱K폫c e0eiP(a,i)=R+KE(i)+c볆쯣돶 ae0ei //R쯹폐폃뮧움럖뗄욽뻹횵 uEnd NewItemRecom() DISCF쯣램쪵퇩짨볆폫뷡맻럖컶 DISCF쯣램뗄쪵퇩짨볆 캪퇩횤DISCF쯣램뗄좫쏦탔쓜ꎬ짨볆쪵퇩쪱믹폚뾼싇틔쿂죽룶랽쏦뗄탨튪ꎺ ꎨ1ꎩ 뇈뷏놾컄쳡돶뗄쯣램폫믹폚떥룶탅쾢풴뗄췆볶쯣램뗄췆볶킧맻뗄폅쇓ꆣ튲뻍쫇짨볆웤폫펦폃맣랺뗄죽룶쾵춳뇈뷏ꎬꋙ SCFꆪ믹폚떥탅쾢풴쿠쯆폃뮧쾵춳ꎻꋚ ECFꆪ믹폚떥탅쾢풴볒폃뮧쾵춳ꎻꋛ HCFꆪ뻟폐좨훘뗄쿠쯆폃뮧뫍볒폃뮧뗄믬뫏췆볶쾵춳ꆣ ꎨ2ꎩ 쯦ퟅ듽췆볶쿮쒿붻뮥탔뗄퓶잿뫍췆볶쇬폲탨튪쏅뗄벼쫵횪쪶뗄퓶볓ꎬ놾컄쳡돶쯣램뗄췆볶킧맻ꆣ ꎨ3ꎩ놾컄쳡돶뗄쯣램퓚뷢뻶쫽뻝쾡쫨탔컊쳢뗄쓜솦죧뫎ꆣ 놾컄뗄럂헦쪵퇩쯹폃믺웷펲볾엤훃쫇ꎺIntel Pentium 4뒦샭웷ꎬ1M쓚듦ꎬ80G펲엌ꆣ퓋탐뮷뺳쫇ꎺ닙ퟷ쾵춳Windows xPꎬ뾪랢욽첨쫇MyEclipseꎬ뇠돌폯퇔캪JAVAꎬ쫽뻝뿢쾵춳캪SQL Server2000ꆣ DISCF쯣램뗄쪵퇩쫽뻝 캪좫쏦움볛DISCF쯣램뗄탔쓜ꎬ컒쏇럖뇰뷸탐솽ퟩ쪵퇩ꆣ ꎨ1ꎩ뺲첬쫽뻝벯뗄풤놸쪵퇩 룃쪵퇩쯹쪹폃뗄쫽뻝벯살ퟔMinnesota 듳톧GroupLens Research 쿮쒿ퟩ쫕벯뗄MovieLens쫽뻝벯ꆣMovieLens햾뗣ꎨhttpꎺ//MovieLensꎮUmnꎮedu/ꎩ쫇튻룶믹폚Web뗄퇐뺿탍췆볶쾵춳ꎬ붨솢폚1977쓪ꎬ쎿탇웚뚼폐짏냙뗄폃뮧럃컊룃쾵춳ꎬ뷸탐뗧펰움볛뫍믱뗃맘폚뗧펰뗄췆볶ꆣ룃쫽뻝벯냼몬943룶폃뮧뛔1682늿뗧24
펰뗄움럖ꎬ놾컄뛔듓MovieLens햾뗣쿂퓘뗄쫽뻝컄볾뷸탐쇋룱쪽뗄뮻ꎬ헻샭뫳떼죫떽SQL SERVER2000쫽뻝뿢훐ퟷ캪쪵퇩쫽뻝ꎬ냼몬쇋100캻폃뮧ꎬ600늿뗧펰ꎬ폃뮧뛔뗧펰뗄움럖캪10022룶ꆣ쫽뻝벯평죽헅쫽뻝뇭ퟩ돉:User(폃뮧뇭)ꆢItem(쿮쒿뇭ꎬ벴뗧펰뇭)ꆢRate(폃뮧움럖뇭)ꎬ룷뇭뗄ퟩ돉죧쿂뇭쯹쪾ꆣ뛸쟒ꎬ냑쪵퇩쫽뻝벯내4뇈1싊럖돉통솷벯뫍닢쫔벯ꆣ폃뮧뛔뗧펰뗄움볛횵듓1떽5ꎬ횵풽룟듺뇭폃뮧뛔룃뗧펰뗄탋좤뛈풽룟ꆣ ꎨ2ꎩ뗧ퟓ짌컱췸햾뗄럂헦쫽뻝쪵퇩 ힿ풽췸짏릺컯ꎨ ꎩ캪퇇십톷웬쿂릫쮾ꎬ돉솢폚2000쓪5퓂ꎬퟜ늿캻폚놱뺩ꆣퟷ캪훐맺뗧ퟓ짌컱쇬탤ꎬힿ풽퇇십톷캪쿻럑헟쳡릩냼삨쫩벮ꆢ틴샖ꆢ틴쿱ꆢ죭볾ꆢ쫽싫ꆢ볒뗧ꆢ췦뻟ꆢ볒뻓뗈뻅듳샠ꎬ뎬맽150췲훖닺욷틔릩톡퓱ꎬ쎿쳬폐돉잧짏췲쿻럑헟춨맽룃췸햾쳥퇩췸짏릺컯뗄샖좤ꆣ 뇭 User(폃뮧뇭) 뇭뷡릹 ퟖ뛎 쮵쏷 UserId 폃뮧뇠뫅 Gender 폃뮧탔뇰 Age 폃뮧쓪쇤 Occupation 폃뮧횰튵 뇭 Item(쿮쒿뇭) 뇭뷡릹 ퟖ뛎 쮵쏷 MovieId 뗧펰뇠뫅 Title 뗧펰쏻돆 Genres 뗧펰럖샠 뇭 Rate(폃뮧움럖뇭) 뇭뷡릹 ퟖ뛎 쮵쏷 UserId 폃뮧뇠뫅 MovieId 뗧펰뇠뫅 Rating 움럖 25
캪퇩횤짌욷뗄붻뮥쳘탔뛔놾컄쯹쳡돶쾵춳뗄췆볶킧맻ꎬ컒쏇톡퓱솽ퟩ쪵퇩쫽뻝ꎬ튻쫇볛룱뷏뗍뗄쫩벮뗄쿺쫛쫽뻝ꎬ듺뇭붻뮥탔뷏죵뗄짌욷ꎬ쿻럑헟퓚톡퓱헢샠짌욷쪱뿉쓜믡듓쿠쯆폃뮧훐믱좡붨틩ꎬ쇭튻룶쫇짌욷볛룱뷏룟뗄뇊볇놾ꎬ듺뇭붻뮥탔잿뗄짌욷ꎬ뛔폚헢샠짌욷뗄톡퓱ꎬ쿻럑헟뿉쓜룼풸틢쳽좡볒폃뮧뗄틢볻ꆣ 컒쏇럖뇰톡퓱500놾늻춬샠뇰뗄춼쫩뫍100훖뇊볇놾뗧쓔폃폚뷓쫕뿍뮧뛔짌욷뗄움럖늢쳡릩쿠펦뗄짌욷췆볶쇐뇭ꎬ늢뛔쳡릩뗄짌욷뷸탐듲럖ꎬ럖횵듓0ꆪ5ꎬ럖뇰뇭쪾죧쿂ꎺ5럇뎣뫃ꎬ4뫜뫃ꎬ3뫃ꎬ2튻냣ꎬ1닮ꎬ0뇭쪾듋짌욷쎻폐죎뫎죋룸돶움싛ꆣ쿂쏦쯦틢헒100룶춬톧ꎬ퓚쯻쏇쫂쿈늻횪뗀쪵퇩쒿뗄뗄쟩뿶쿂ꎬ룸쯹쳡릩뗄짌욷뷸탐듲럖ꎬ뗃돶튻ퟩ쫽뻝벯ꎬ죧춼쯹쪾ꎺ 춼ꎺ춼쫩뗄쒣쓢쫽뻝벯 죧짏춼쯹쪾ꎬ춼짏쯹쪾쫽뻝랴펳뗄쫇뇠뫅캪1ꆢ2ꆢ3뗄톧짺뛔춼쫩짌욷뗄움럖횵ꆣ벴듓ퟳ훁폒럖뇰듺뇭:톧짺IDꆢ춼쫩짌욷ID틔벰톧짺뗄움럖탋좤뛈ꎬ죧쫽뻝벯훐뗚튻탐쫽뻝뇭쏷톧짺1뛔뇠뫅캪10춼쫩뗄탋좤뛈캪0ꆣ춬짏쯹쫶ꎬ100쏻춬톧뛔뇊볇놾뗄탋좤뛈죧춼쯹쪾ꆣ떫쫇헢솽룶쫽뻝벯폐튻룶ퟮ듳뗄좱뗣뻍쫇움럖쫽뻝쾡쫨탔ꎬ뛔솽룶쫽뻝벯ꎬ샻폃뗄랽램뷸탐쇋풤뒦샭ꎬ뒦샭뫳뗄뷡맻ퟷ캪쪵퇩뗄쫤죫쫽뻝ꆣ 26
DISCF쯣램뗄쪵퇩움볛뇪ힼ 죧뫎움볛췆볶쾵춳쫇럇뎣훘튪뗄컊쳢ꎬ짨볆뫃뗄움볛횸뇪쳥쾵튲쫇쯣램훐훘튪뮷뷚ꆣ쒿잰펦폃췆볶쾵춳훐ퟮ맣랺움볛횸뇪뿉틔럖돉솽룶랽쏦ꎺ튻쫇춳볆뗄ힼ좷탔ꎻ뛾쫇췆볶뷡맻럖샠뗄ힼ좷뛈ꆣ 춳볆뗄ힼ좷탔횸뇪럖돉솽룶늿럖 ꎨ1ꎩ뢲룇싊 폃뮧움맀췆볶쾵춳뿉틔쳡릩뗄췆볶햼쯹폐쿮쒿뗄뇈샽ꆣ 춼 뇊볇놾뗄쒣쓢쫽뻝벯 ꎨ2ꎩ춳볆ힼ좷탔 춳볆ힼ좷탔횸뇪춨맽뇈뷏듽췆볶폃뮧움맀돶뗄췆볶럖횵폫듽췆볶폃뮧뗄쪵볊움럖횵뷸탐뇈뷏ꆣ욽뻹뻸뛔욫닮(MeanAbsoluteErmrꎬMAE)쫇맣랺펦폃뗄움볛랽램ꆣ뛔폚쎿뛔풤닢움럖폫쪵볊움럖쫓캪튻쿮ꆣ쫗쿈볆쯣쎿튻룶쿮쒿헢솽룶횵뗄닮뗄뻸뛔횵ꎬ좻뫳붫헢킩쫽횵샛볓ꎬퟮ뫳돽틔쿮쒿뗄쫽쒿ꆣ웤릫쪽죧쿂ꎺ p−r∑ijijMAE= N웤훐ꎬN듺뇭쯹폐뗄폃뮧움럖ꎬp듺뇭폃뮧i뛔쿮쒿j뗄풤닢움럖ꎬrijij27
듺뇭i뛔j뗄쪵볊뗄움럖ꆣ볆쯣돶뗄MAE횵풽킡ꎬ횤쏷췆볶쾵춳뗄뿉뾿탔풽잿ꆣ 췆볶뷡맻럖샠뗄ힼ좷뛈 뇭쪾췆볶뷡맻움볛뇪ힼꎬ벴췆볶쿮쒿폫폃뮧탋좤욥엤돌뛈ꎬ샽죧튻룶폃뮧뛔닺욷뗄움럖뎬맽ퟮ룟움럖뗄70%(샽죧ퟮ룟움럖캪5ꎬ움럖캪3)믲헟릺싲쇋헢룶닺욷ꎬ뻍죏캪폃뮧뛔헢룶쿮쒿룐탋좤ꆣ죴췆볶쾵춳뛔짌욷뗄움럖듳폚ퟮ룟움럖뗄70%ꎬ뻍믡냑헢룶짌욷췆볶룸쒿뇪폃뮧ꆣ웤춨뎣폃뺫좷싊(precision) 뫍헙믘싊(Recall)쮵쏷ꎬ솽헟횮볤뗄몬틥죧뇭 뇭ꎺ 럖샠ힼ좷뛈쪾틢뇭 쪵볊쾲뮶 늻쾲뮶 췆볶쿮쒿AB캴췆볶쿮쒿 C D 췆볶쾵춳룸쒿뇪폃뮧췆볶뗄짌욷훐ꎬ헦헽쿫튪뗄ꆢ튲뻍쫇쿠맘뗄풽뛠풽뫃ꎬ늻쿠맘뗄풽짙풽뫃ꎬ헢뻍쫇ꆰ헙믘싊ꆱꎬ벴퓚뇭훐ꎺRecall = A/(A+B)ꆣ춬퇹ꎬ컒쏇폃뺫좷싊(precision)뇭쪾쾵춳췆볶룸폃뮧룐탋좤뗄짌욷햼폃뮧쪵볊쾲뮶뗄ퟜ짌욷쫽솿뗄뇈샽ꎬ벴A/(A+C)ꎬ풽듳풽뫃ꆣꆰ헙믘싊ꆱ폫ꆰ뺫좷싊ꆱ쯤좻쎻폐뇘좻뗄맘쾵ꎨ듓짏쏦릫쪽훐뿉틔뾴떽ꎩꎬ떫퓚쪵볊펦폃훐ꎬ쫇쿠뮥훆풼뗄ꆣ튪룹뻝쪵볊탨쟳ꎬ헒떽튻룶욽뫢뗣ꆣퟛ뫏뾼싇ꎬ폖폐쇋F1움볛뇪ힼꎬ죧릫쪽쯹쪾ꆣ F1 =(2 *precision * recall)/(precision + recall) DISCF쯣램뗄쪵퇩맽돌뫍뷡맻럖컶 헻룶쪵퇩럖돉죽룶늿럖ꎺ 쪵퇩튻뷨훺MovieLens 뺲첬쫽뻝벯살에뛏DISCF쯣램쾵춳폫웤쯻죽룶췆볶쯣램쾵춳퓚MAE뫍뢲룇싊뗄탔쓜뇈뷏ꎬ쪵퇩뷡맻죧뇭ꆣ 뇭ꎺDISCF쾵춳폫죽훖뗤탍뗄킭춬맽싋쯣램춳볆ힼ좷탔뇈뷏 쾵춳 MAE 뢲룇싊ꎨ%ꎩ SCF ꎨ%ꎩ ꎨ%ꎩ ECF ꎨ%ꎩ ꎨ%ꎩ HCF ꎨ%ꎩ ꎨ%ꎩ 퓚뇭훐쿔쪾ퟮ훕뗄쪵퇩뷡맻ꎺ놾컄쳡돶뗄DISCF쯣램쾵춳뗄MAE횵럖뇰28
뛔펦SCFꆢECF뫍HCF쾵춳뗍폚ꆢ뫍룶냙럖뗣ꎬ쫽횵늻쫇뫜샭쿫ꎬ떫쮵쏷DISCF쯣램쾵춳뿉뾿탔ꎬ쯦ퟅ듽풤닢쿮쒿붻뮥탔쳡룟뫍움럖쫽뻝쾡쫨탔퓶볓ꎬDISCF쯣램쾵춳뗄폅쫆믡룼쏷쿔ꆣ퓚풤닢뗄뢲룇싊랽쏦ꎬSCF쾵춳뇈DISCF쯣램쾵춳뗍룶냙럖뗣ꎬ뇈HCF쾵춳룟룶냙럖뗣ꆣ 쪵퇩뛾뗄쫽뻝풴살ퟔ컒쏇틑뒦샭뫃뗄ힿ풽췸짏릺컯ꎨ ꎩ뗧ퟓ짌컱췸햾럂헦쫽뻝ꎬ쒿뗄움볛DISCF쯣램쾵춳폫웤쯻죽룶췆볶쯣램쾵춳퓚췆볶ힼ좷탔탔쓜횸뇪뺫좷싊(precision)ꆢ헙믘싊(Recall)뫍F1ꆣ쪵퇩뷡맻죧뇭쯹쪾ꆣ 뇭ꎺDISCF쾵춳폫죽훖뗤탍뗄킭춬맽싋쯣램췆볶ힼ좷탔뇈뷏 Precision recall F1쾵춳 뇊볇놾 춼쫩 뇊볇놾 춼쫩 뇊볇놾 춼쫩 DISCF SCF ECF HCF 듓뇭훐뾴떽ꎬ퓚F1횸뇪훐ꎬDISCF쾵춳뇈SCF쾵춳퓚뇊볇놾뗄췆볶훐쳡룟쇋34%ꎬ떫퓚헫뛔춼쫩뗄췆볶훐횻쳡룟쇋%ꎬ헢튲횤쏷쇋쯦ퟅ짌욷뗄붻뮥탔뗄퓶볓ꎬDISCF쾵춳뇭쿖돶살뗄폅쫆풽랢쏷쿔ꆣ컒쏇퓚쪵퇩횮돵믡죏캪ꎬ쯦ퟅ짌욷붻뮥탔뗄퓶볓ꎬ죋쏇퓚릺싲뇊볇놾쪱ꎬ믡룼쿠탅볒폃뮧뗄췆볶ꎬ떫쪵퇩뷡맻잡잡쿠랴ꎬ뇊볇놾짌욷뗄췆볶훐ꎬDISCF쾵춳튪뇈ECF쾵춳뇭쿖룼돶즫ꎬ풭틲튲탭쫇컒쏇쪵퇩쫽뻝벯뫏첫짙ꎬ뛸쟒듓톧짺훐톡퓱늿럖폃뮧ퟷ캪볒폃뮧ꎬ헢킩볒폃뮧뗄쮵럾솦늻릻퓬돉뗄뫳맻ꆣ 쪵퇩죽에뛏DISCF쾵춳퓚뷢뻶쫽뻝쾡쫨탔랽쏦뗄쓜솦ꆣ킭춬맽싋쾵춳뗄탔쓜좡뻶폚뿉뾿뗄움럖뻘헳ꎬ뎣폃뗄킭춬맽싋쯣램뚼쏦쇙쫽뻝쾡쫨뗄컊쳢ꎬ쯦ퟅ쫽뻝쾡쫨탔뗄퓶볓ꎬ쯣램뗄췆볶킧맻벱뻧쿂붵ꆣ캪퇩횤DISCF쾵춳퓚뷢뻶쫽뻝쾡쫨탔뗄컊쳢ꎬ컒쏇쪵퇩퇩횤퓚늻춬뗄쫽뻝쾡쫨탔쿂DISCF쾵춳폫SCF쾵춳횮볤췆볶탔쓜뗄폅쇓ꆣ 퓚풭쪼쫽뻝훐ꎬ100쏻톧짺뛔100훖뇊볇놾뷶폐786룶움럖쫽쒿ꎬ웤쫽뻝쾡쫨뛈캪1-786/100*100=ꎬ춼쫩뗄풭쪼쫽뻝뗄쾡쫨뛈캪1-8893/100*500=ꎬ춬퇹내헕쪵퇩뛾뗄뇪ힼ냑풭쪼쫽뻝럖돉통솷벯뫍닢쫔벯ꎬ떫퓚쫤죫통솷벯쪱ꎬ캪쮵쏷쫽뻝뗄쾡쫨뛈ꎬ컒쏇냑통솷벯뗄100%ꆢ88%ꆢ75%ꆢ63%ꆢ50%ꆢ38%뫍25%럖쏷쫤죫ꎬ살닢쫔DISCF쾵춳폫SCF쾵춳횮볤췆볶탔쓜뗄폅쇓ꆣ 헫뛔붻뮥탔잿뗄뇊볇놾뗄췆볶킧맻죧춼쯹쪾ꎬ쯦ퟅ쫽뻝쾡쫨뛈뗄퓶볓솽룶쾵춳뗄췆볶쓜솦훰붥복죵ꎬDISCF쾵춳뇈SCF쾵춳폅쫆럹뛈퓚%떽29
%횮볤ꎬ떫뛔폚붻뮥탔죵뗄춼쫩뗄췆볶ꎬDISCF쾵춳뇈SCF쾵춳폅쫆럹뛈퓚%떽%횮볤ꎬ췆볶킧맻죧춼쯹쪾ꆣ DISCFSCF(D-S)/쫽뻝뗄쾡쫨뛈 춼ꎺ늻춬쫽뻝쾡쫨뛈붻뮥탔잿뗄뇊볇놾췆볶킧맻 DISCFSCF(D-S)/쫽뻝뗄쾡쫨뛈 춼ꎺ늻춬쫽뻝쾡쫨뛈붻뮥탔죵뗄춼쫩췆볶킧맻 춨맽뛔뇈쪵퇩럖컶ꎬ컒쏇뿉틔뾴떽쫽뻝쾡쫨탔룟ꎬDISCF쾵춳뇈SCF쾵춳뗄췆볶킧맻풽뫃ꎬ헢쮵쏷놾컄짨볆뗄DISCF쾵춳퓚뷢뻶쫽뻝뗄쾡쫨탔컊쳢훐뇭쿖늻듭ꎬ좻뛸퓚붻뮥탔죵뗄춼쫩뗄췆볶훐ꎬSCF쾵춳뇈DISCF쾵춳룼죝틗쫜떽쫽뻝쾡쫨탔뗄펰쿬ꎬ헢튲쮵쏷쇋퓚붻뮥탔잿뗄뇊볇놾췆볶훐ꎬ쿻럑헟룼쟣쿲폚뷓쫜볒뗄틢볻ꆣ떫쫇ꎬ컒쏇뗄쪵퇩뷡맻살ퟔ폚쳘쫢뗄뗧ퟓ짌컱뮷뺳뫍쳘쫢뗄쪵퇩쫽뻝벯뫏ꎬ컒쏇탨튪퓚룼뛠뗄뗧ퟓ짌컱욽첨짏살퇩횤DISCF쾵춳ꆣ 놾헂킡뷡 놾헂럖컶DISCF쯣램뗄쳘뗣ꎬ뷩짜웤쪵쿖뗄맽돌ꆣ캪퇩횤DISCF쾵춳뗄탔쓜ꎬ컒쏇뷨훺뺲첬쫽뻝뫍뗧ퟓ짌컱췸햾뗄럂헦쫽뻝ꎬ짨볆죽룶쪵퇩욽첨ꎬ뷡맻뇭쏷ꎬDISCF쯣램쾵춳뇈뒫춳뗄킭춬맽싋쯣램퓚MAE횵뫍풤닢뗄뢲룇싊랽쏦뇭쿖폅풽ꎬ퓚췆볶뗄ힼ좷탔랽쏦춬퇹잿폚뒫춳뗄킭춬맽싋쾵춳ꎬ뛸쟒퓚쏦뛔쫽뻝30 F1F1폫탔쓜뇈뷏DISCFSCF폫탔쓜뇈뷏DISCFSCF
쾡쫨탔쟩뿶쿂뇭쿖돶뷏뫃뗄췆볶킧맻ꆣ떫쫇쪵퇩뷶뷶퓚럂헦쪵퇩믹뒡짏뗃떽퇩횤ꎬ컒쏇뷸튻늽뗄릤ퟷ붫냑룃쾵춳퓚늻춬뗄쪵볊뗄뗧ퟓ짌컱췸햾퓋탐ꎬ퓚췆볶쿮쒿붻뮥탔쳡룟뗄쟩뿶쿂ꎬ늻뛏룄뷸쯣램ꎬ쪹췆볶쾵춳뗄뿉뾿탔룼룟ꆣ31
뗚컥헂 ퟜ뷡뫍햹췻 뗧ퟓ짌컱엮늪랢햹첬쫆쿂ꎬ췆볶쾵춳틑돉캪룶탔뮯뗄뗧ퟓ짌컱췸햾뗄훘튪ퟩ돉늿럖뫍쳡룟뗧ퟓ짌컱럾컱짌뺺헹솦뗄횧독솦솿ꆣ킭춬맽싋ꎨCFꎩퟷ캪췆볶쾵춳훐ퟮ돉릦뗄튻쿮벼쫵ꎬ쯼뗄쓜솦틑뺭퓚늻춬뗄뗧ퟓ짌컱쾵춳훐뗃떽쇋퇩횤ꆣ 좻뛸ꎬ뒫춳CF쏦쇙탂뗄쳴햽ꆪ움럖쫽뻝뗄룟캬쾡쫨탔컊쳢ꆢ탂폃뮧컊쳢ꆢ췆볶뷡맻뗄뿉뾿탔컊쳢뗈뗈ꆣ폈웤쫇CF퓚헫뛔붻뮥탔잿ꆢ탨튪쏅벼쓜놳뺰횪쪶뗄닺욷뫍럾컱뗄췆볶ꎬ룼쿔뗃솦늻듓탄ꆣ쏦뛔튻쾵쇐뗄벬쫖뗄컊쳢ꎬ퇐뺿헟틑쳡돶쇋CF폫믹폚쓚죝맽싋ꎨcontent-based filteringꎩ뷡뫏뗄믬뫏췆볶벼쫵ꆣ떫쫇ꎬ믹폚쓚죝맽싋뗄쯣램뇘탨쫗쿈돩좡닺욷쫴탔쳘헷ꎬ듋쿮릤ퟷ랴뛸퓶볓쇋췆볶쾵춳뗄뢺떣ꆣ캪쇋뷢뻶킭춬맽싋쯣램훐폶떽뗄쫽뻝쾡쫨탔ꆢ헫뛔뻟폐쏅횪쪶놳뺰뗄닺욷췆볶ꎬ놾컄쳡돶믹폚쮫탅쾢풴쒣쪽뗄킭춬맽싋쯣램ꎨDISCF쯣램ꎩꆣ 놾컄훷튪릤ퟷퟜ뷡 놾컄쯹ퟶ뗄훷튪퇐뺿릤ퟷ죧쿂ꎺ ꎨ1ꎩ듓뗧ퟓ짌컱훐뗄췆볶쾵춳죫쫖ꎬ닩퓄쇋맺쓚췢듳솿쿠맘쇏ꎬ룅쫶쇋뗧ퟓ짌컱훐췆볶쾵춳뗄ퟷ폃뫍펦폃쪵샽ꎻ룸돶뗧ퟓ짌컱췆볶쾵춳뗄쒣탍ꎬ늢뷩짜쇋췆볶쾵춳훐뗄쫤죫쫽뻝ꎻ뛔뗧ퟓ짌컱훐뗄췆볶쾵춳뗄럖샠뫍펦폃뗄뗤탍벼쫵ퟶ쇋뷏캪짮죫뗄퇐뺿ꆣ ꎨ2ꎩ뛔킭춬맽싋뗄췆볶쯣램뷸탐퇐뺿럖컶ꎬ쮵쏷킭춬맽싋쯣램뗄릤ퟷ풭샭뫍쯣램뗄쫤죫쫤돶ꎻ뷩짜쇋퓚췆볶쾵춳훐펦폃쪮럖웕뇩뗄솽훖킭춬맽싋쯣램ꎺ튻훖쫇믹폚폃뮧뗄킭춬맽싋쯣램ꎬ쇭튻훖쫇믹폚쿮쒿뗄킭춬맽싋쯣램ꎻ훘뗣쳖싛쇋뒫춳뗄킭춬맽싋쯣램듦퓚뗄컊쳢ꎬ늢럖컶쇋떱잰쳡돶뗄뷢뻶컊쳢뗄랽램ꎬ횸돶쯼쏇뗄폅쫆뫍늻ퟣ횮뒦ꆣ ꎨ3ꎩ캪쇋뷢뻶킭춬맽싋쯣램훐폶떽뗄쫽뻝쾡쫨탔ꆢ헫뛔뻟폐쏅횪쪶놳뺰뗄닺욷췆볶ꎬ싛컄쳡돶쇋믹폚쮫탅쾢풴쒣쪽뗄킭춬맽싋쯣램ꎨDISCF쯣램ꎩꎬ룃쯣램에뛏쒿뇪뿍뮧뛔믮뚯쿮쒿뗄탋좤돌뛈좡뻶폚솽룶췆볶ퟩꆪꆪ쿠쯆폃뮧췆볶ퟩꎨퟮ뷼쇚뻓벯뫏ꎩ폫볒췆볶ퟩꎬ냑솽룶췆볶ퟩ뗄움럖뇭룱뷡뫏웰살ꎬퟩ돉탅죎뛈룟뗄탅쾢풴ꎬ뷓ퟅꎬ쳖싛룷ퟔ펰쿬쒿뇪뿍뮧뛔믮뚯쿮쒿뗄좨훘ꎬ볆쯣돶쒿뇪뿍뮧뗄ퟮ훕탋좤뛈ꎬ쪵쿖쾵춳췆볶ꆣ믹폚쮫탅쾢풴쒣쪽뗄킭춬맽싋쯣램ꎨDISCFꎩ돤럖뾼싇뗧ퟓ짌컱훐룶탔뮯럾컱뗄쪵볊쟩뿶ꎬ쪹췆볶쾵춳붨솢퓚룼뿉뾿뗄탅쾢풴믹뒡짏ꎬ쫔퇩횤쏷룃랽램폐룼뫃뗄췆볶훊솿ꆣ 뷸튻늽뗄퇐뺿릤ퟷ 뺡맜놾죋퓚럖컶뒫춳킭춬맽싋쯣램믹뒡짏ꎬ쳡돶믹폚쮫탅쾢풴쒣쪽뗄킭춬맽32
싋쯣램ꎨDISCFꎩꎬ늢쟒좡뗃쇋튻뚨뗄돉맻ꎬ떫쫇컞싛쫇쯣램놾짭ꎬ쳘뇰쫇쪵볊펦폃ꎬ뚼폐듽뷸튻늽뗄삩햹뫍짮죫퇐뺿ꆣ퓚놾컄릤ퟷ뗄믹뒡짏ꎬ붫듓틔쿂랽쏦햹뾪뷸튻늽뗄퇐뺿릤ퟷꎺ ꎨ1ꎩ듳탍뗧ퟓ짌컱췆볶쾵춳뇘탫쓜릻춬쪱캪쫽틔췲볆뗄뿍뮧닺짺쪵쪱뗄췆볶쇐뇭ꎬ쯦ퟅ뿍뮧쫽솿뫍짌욷쫽솿뗄늻뛏퓶볓ꎬ뗧ퟓ짌컱췆볶쾵춳뗄짬쯵쓜솦뫍쪵쪱탔튪쟳풽살풽쓑틔놣횤ꎬ탨튪냑믹폚쮫탅쾢풴쒣쪽뗄킭춬맽싋쯣램ꎨDISCFꎩ퓚쪵쿖짏럖돉솽룶늿럖ꆪ샫쿟뗄쫽뻝뒦샭뫍퓚쿟뗄쪵쪱췆볶ꆣ ꎨ2ꎩ믹폚쮫탅쾢풴쒣쪽뗄킭춬맽싋쯣램ꎨDISCFꎩ훐탨튪볆쯣솽룶탅쾢풴뗄뿉탅뛈ꎬ평폚늻춬뗄폃뮧뛔솽룶탅쾢풴뗄뿉탅뛈뗄뛠퇹탔ꎬ튪쟳쳡돶룼뫃뗄랽램냑뿉탅뛈뗄볆쯣붨솢퓚룼뫏샭뗄믹뒡짏ꆣ ꎨ3ꎩ킭춬맽싋벼쫵훕벫쒿뇪쫇쿲뿍뮧쳡릩룟훊솿ꆢ뿉뾿뗄닺욷탅쾢럾컱췆볶ꎬ횻폐냑쯼뫍웤쯼탅쾢맽싋랽램ꆢ탅쾢췆볶벼쫵폐믺뷡뫏ꎬ닅쓜릻폐룼뫃뗄킧맻ꆣ샽죧ꎺ뿉틔뷨본볒쾵춳훐죋믺붻뮥뗄돉릦낸샽ꎬ죋믺붻뮥쒣쪽펦폃떽DISCF쯣램훐ꎺ냑췆볶맽돌뗄횱맛뷢쫍쳡릩룸뿍뮧ꎬ죃뿍뮧닎폫헻룶췆볶맽돌늢ퟶ돶쿠펦뗄뻶닟ꎬ살쳡룟췆볶뷡맻뗄뿉뾿탔ꆣ쿖퓚폯틥web벼쫵뗄늻뛏돉쫬ꎬ킭춬맽싋벼쫵폫횮뷡뫏ꎬ돤럖랢믓폯틥탅쾢뗄ퟷ폃ꎬ룼뫃뗘웋ힽ뿍뮧뚯첬탋좤잨틆쒣쪽ꎬ듳듳쳡룟췆볶쾵춳뗄췆볶훊솿ꆣ 33
닎뾼컄쿗 [1] Resnick and VarianꎮRecommender systems[J].Communications of the ACMꎻ1997ꎬ40 (3)ꎺ56 – 58ꎮ [2] 폠솦ꎬ쇵슳ꎮ뗧ퟓ짌컱룶탔뮯췆볶퇐뺿[J].볆쯣믺벯돉훆퓬쾵춳ꎬ2004ꎬ10(10)ꎺ1306ꆪ1313 [3] MARKO BꎬYOAV SꎮFAB ꎺ content-based collaborative recommendation [J].Communications of the ACMꎬ1997 ꎬ40 (3) ꎺ66 - 72ꎮ [4] GOOD NꎬSCHAFER J BꎬKONSTAN Jꎬet alꎮCombining collaborative filtering with personal agents for better recommendations[A].Proceedings of AAAI'99[C].1999ꎮ [5] 쇎쿾ꎮ훐맺뗧ퟓ짌컱놨룦[M].놱뺩ꎺ뺭볃뿆톧돶냦짧ꎬ2005 [6] 킻훐ꎮ믹폚Web 쫽뻝췚뻲짌컱췸햾췆볶쾵춳뗄퇐뺿[쮶쪿톧캻싛컄]ꎮ훘쟬ꎺ컷쓏쪦랶듳톧ꎬ2002 [7] S~araꎬKar IPisGꎬKonstanJꎬetalꎮItem튻based collaborative filtering reoommendation algorithms[C].proceeding of the10 th Intemational WWW Conference NewYerkꎺACM Pressꎬ2001ꎺ285튻295 [8] 헅몣퇠ꎬ뚡럥ꎬ붪샶뫬ꎮ믹폚쒣뫽뻛샠뗄킭춬맽싋췆볶랽램[J]. 볆쯣믺럂헦ꎬ2005ꎬ22(8):144-147 [9] Mobasher BꎬJi nxꎬzhou YꎮSemantieally enhanced collaborative f iitering on the web[C].proceedings of the European Web Mining Forum ResnikꎺSpringer Pressꎬ2004ꎺ57튻76ꎮ [10] 췵쾼ꎬ쇵쟙ꎮ킭춬맽싋퓚췆볶쾵춳훐뗄펦폃퇐뺿[J].볆쯣믺쾵춳펦폃ꎬ2004(4) [11] 퇮쏷뮨ꎬ맅횾쏱ꎮ 믹폚뺭퇩뗄킭춬췆볶[J]. 킡탍캢탍볆쯣믺쾵춳ꎬ2008ꎬ29(3)ꎺ498ꆪ502 [12] BꎮJoseph Pine II 1993ꎮMass CustomizationꎮHarvard Business School PressꎮBostonꎬMassachusetts [13] HuangꎬZꎮꎬ Wꎮ Chungꎬ et alꎬꆰA graph model forE-Commerce recommender systemsꆱꎬJournal of theAmerican Society for Information Science andTechnologyꎬVolꎮ55ꎮNoꎮ3ꎬ pp259-74ꎬ2004ꎮ [14] ScharerꎬJꎮBꎬ JꎮAꎮKonstanꎬet alꎬꆰE-commerce recommendation applicationsꆱꎬData Mining and Knowledge DiscoveryꎬVolꎮ5ꎬNoꎮ1ꎬpp115-53ꎬ2001ꎮ [15] Resnick Pꎬ Iacovou Nꎬet alꎬꆰGrouplensꎺan open architecture for collaborative filtering of netnewsꆱꎬProceedings of the ACM Conference on Computer Supported Cooperative Workꎬpp175-86ꎬ1994ꎮ 34
[16] ShapiraꎬBꎬ MꎮTaieb-Maimonꎬet alꎬꆰStudy of the usefulness of known and new implicit indicators and their optimal combination for accurate inference of users interestsꆱꎬProceeding of ACM Symposium on Applied Computingꎬpp1118-19ꎬ2006ꎮ [17] ScharerꎬJꎮBꎬ JꎮAꎮKonstanꎬ Riedl J ꆰrecommender system in E-commerceꆱIn ACM Conference on Electronic Commerce(EC99)ꎬ1999ꎮ [18] AdomaviciusꎬGꎮꎬAꎮTuzhilinꎬ ꆰToward the next generation of recommender systemꎺ a survey of the state-of-the-art and possible extensionsꆱꎬIEEE Transactions on Knowledge and Data EngineeringꎬVolꎮ17ꎬNOꎮ6ꎬ pp734-749ꎬ 2005ꎮ [19] Zan HuangꎬWingyan Chunꎬand Hsinchun ChenꎮA Graph Model for E-commerce Recommender SystemsꎮInꎺJournal of the American Society for Information Science and Technoloygꎬ55(3)ꎺ259ꆪ274ꎬ2004ꎮ [20] MelvilleꎬPꎬ RꎮJꎮMooneyꎬet alꎬꆰContent-boosted collaborative filtering for improved recommendationsꆱꎬProceeding of Conference on Artificial Intelligenceꎬpp187-92ꎬEdmontonꎬ2002ꎮ [21] ClaypoolꎬMꎮꎬet alꎬꆰCombining content-based and collaborative filters in an online newspaperꆱꎬProceeding of ACM Workshop Recommender SystemsꎺAlgorithms and Evaluationꎬ1999ꎮ [22] Ben S J ꎬKonstan J A ꎬRiedl J ꎮE - commerce Recommendation Applications[ EB/ OL ] ꎮhttp ꎺ//wwwꎮGrouplensꎮorg/ papers/ pdf/ ECRAꎮpdf ꎬ2002 - 09 - 26ꎮ [23] Breese J SꎬHeckerman DꎬKadie CꎮEmpirical Analysis of Predictive Algorithms for Collaborative Filtering[A].In ꎺProceedings of the 14th Conference on Uncertaint y in Artificial Intelligence[C].Madisonꎬ WisconsinꎺMor gan Kaufmannꎬ1998ꎮ43 - 52ꎮ [24] Badrul Sꎬ Karypis Gꎬ Konstan Jꎬet al ꎮItem - based Collaborative Filtering Recommendation Algorithms[Z]. WWW10ꎬ2001ꎮ [25] 뗋낮쇖ꎬ훬퇯폂ꎬ쪩늮샖ꎮ믹폚쿮쒿움럖풤닢뗄킭춬맽싋췆볶쯣램[J]ꎮ죭볾톧놨ꎬ2003ꎬ14 (9) ꎺ1621 - 1628ꎮ [26] Badrul Sꎬ Karypis Gꎬ Konstan Jꎬet alꎮAnalysis of Recommendation Algorithms for E – commerce[M]. New YorkꎬUSAꎺ ACM Pressꎬ 2000ꎮ106 - 112ꎮ [27] 헔쇁ꎬ뫺쓋뺲ꎬ 헅쫘횾ꎮ룶탔뮯췆볶쯣램짨볆[J].볆쯣믺퇐뺿폫랢햹ꎬ 2002ꎬ 39 (8)ꎺ986 - 991ꎮ [28] 뗋낮쇖ꎬퟳퟓ튶ꎬ훬퇯폂ꎮ믹폚쿮쒿뻛샠뗄킭춬맽싋췆볶쯣램[J]. 35
킡탍볆쯣믺쾵춳ꎬ2004 ꎬ25 (9) ꎺ1665 - 1670ꎮ [29] 폎컄ꎬ튶쮮짺ꎮ 뗧ퟓ짌컱췆볶쾵춳훐뗄킭춬맽싋췆볶[J]. 볆쯣믺벼쫵폫랢햹ꎬ2006ꎬ16 (9)ꎺ70 - 72ꎮ [30] 닌뗇ꎬ슬퓶쿩ꎬ샮퇜듯ꎮ 탅쾢킭춬맽싋[J]. 볆쯣믺뿆톧ꎬ2002ꎬ29 (6)ꎺ1 - 4ꎮ [31] Dingꎬ C H Qꎮ A similarityꆪbase probality model for latent semantic indexing[ M]ꎮ Inꎺ proc of the SIGIR ACM 1999ꎮ [32] LH Hꎬ Lu ZꎬLi Yꎮ EVSꎺEnhanced Vector Similarity for collaboratice Information recommendation base on SVD Inꎺ INTL ICSC NAISO Congress on comprtational Intelligenceꎺmethods And Application(CIMA 2001)ꎬ2001 [33] Shapira Bꎬ Shoval PꎬHanani Uꎮ Experimentation with an information filtering system that combines cognitive and sociological filtering inetergrated with user stereotypes Decision support system 1999ꎬ27ꎺ5-24ꎮ [34] MARKO Bꎬ YOAV SꎮFABꎺcontent - based collaborative recommendation[J]. Communications of the ACMꎬ1997ꎬ40 (3)ꎺ66 - 72ꎮ [35] ADOMAVICIUS GTUZHIL IN Aꎮ Extending recommendersystemsꎺ a multidimensional approach [ A ]. Proceedings of theFifth ACMConference on Digital Libraries [C]. 2001ꎮ 115 -123ꎮ [36] BꎮSarwar et alꎬꆰAnalysis of Recommendation Algorithms for E-commerceꎬꆱProcꎮ2nd ACM Confꎮ Electronic Commerce (EC 00)ꎬ ACM Pressꎬ2000ꎬppꎮ158–167ꎮ [37] GꎮAdomavicius and AꎮTuzhilinꎬꆰToward the Next Generation of Recommender SystemsꎺA Survey of the State-of-the-Art and Possible ExtensionsꎬꆱIEEE TransꎮKnowledge and Data Engꎮꎬvolꎮ17ꎬnoꎮ6ꎬ2005ꎬ ppꎮ734–749ꎮ [38] RꎮGreinerꎬXꎮSuꎬBꎮShenꎬand WꎮZhouꎮStructural Extension to Logistic RegressionꎺDiscriminative Parameter Learning of Belief Net Classifiersꎬ Machine Learningꎬ59(3)ꎬppꎮ297 - 322ꎬ2005ꎮ [39] PꎮMelvilleꎬRꎮJꎮMooneyꎬand RꎮNagarajanꎮ Content-Boosted Collaborative Filtering for Improved RecommendationsꎬProceedings of the 18th National Conference of Artificial Intelligenceꎬ2002ꎮ [40] DꎮFꎮDuhan et alꎮꆰInfluences on Consumer Use of Word-of-Mouth Recommendation SourcesꎬꆱJꎮ Academy of Marketing Scienceꎬ volꎮ25ꎬ Fall 1997ꎬ ppꎮ283–295ꎮ [41] JꎮLꎮHerlocker et alꎮꆰEvaluating Collaborative Filtering Recommender 36
Systemsꆱ ACM TransꎮInformation Systemsꎬvolꎮ22ꎬnoꎮ1ꎬ2004ꎬ ppꎮ5–53ꎮ [42] TꎮSꎮRobertsonꎬJꎮZielinskiꎬand SꎮWardꎬConsumer BehaviorꎬScottꎬ Foresman and Coꎮ1984ꎮ [43] JꎮOꆯDonovan and BꎮSmythꎬꆰTrust in Recommender SystemsꆱProcꎮ10th Intꆯl ConfꎮIntelligent User Interfaces (IUI 05)ꎬACM Pressꎬ2005ꎬppꎮ167–174ꎮ 37
릥뛁쮶쪿톧캻웚볤랢뇭톧쫵싛컄쟩뿶 1ꎮ뚭좫뗂ꎮ튻훖탂뗄킭춬맽싋쯣램퇐뺿ꎮ CACISꆤ2009톧쫵싛컄벯ꎮ 2ꎮ뚭좫뗂ꎮ폃뮧탋좤잨틆쒣쪽폫룶탔뮯럾컱ꎮ뗧쓔횪쪶폫벼쫵ꎮ2007ꎬ3 ꎨ5ꎩꎮ1280-1286ꎮ 3ꎮ뚭좫뗂ꎮ믹폚쮫탅쾢풴뗄킭춬맽싋쯣램퇐뺿[J]. 뫏럊릤튵듳톧톧놨 2010ꎨ7ꎩ 38
훂 킻 퓚놾컄췪돉횮볊ꎬ컒튪쿲쯹폐룸폨맽컒횸떼ꆢ맘뮳ꆢ횧돖뫍냯훺뗄샏쪦ꆢ쇬떼뫍쟗죋훂틔훔탄뗄룐킻! 쫗쿈튪룐킻컒뗄떼쪦췵뫆뷌쫚ꎬ퓚퇐뺿짺웚볤뗄톧쾰뫍릤ퟷ훐ꎬ췵샏쪦룸폨쇋컒쾤탄횸떼ꎬ쯻퇏뷷쟳쪵뗄훎톧ퟷ럧ꆢ튻쮿늻릶뗄뺴튵뺫짱틔벰헽횱뗄뒦쫀풭퓲뛔컒펰쿬짮뿌ꎬ붫쪹컒훕짺쫜틦ꆣ퓚싛컄뗄톡쳢ꆢ뾪쳢ꆢ탞룄ꆢ뚨룥뫍싛컄뗄탎돉맽돌훐ꎬ췵샏쪦쟣힢쇋듳솿뗄탄톪ꎬ컒뗄쎿튻뗣뷸늽뚼쓽뻛쇋췵샏쪦뗄탁쟚샍뚯ꆣ 룐킻쯞훝톧풺탅쾢릤돌톧풺뗄샮뫨풺뎤ꎬ샮풺뎤퓚컒뿎쳢뗄퇐뺿뾪랢맽돌훐ꎬ쳡돶쇋뫜뛠뒴탂탔뗄틢볻ꎬ돎쟥쇋컒죏쪶샭뷢짏뗄컳쟸ꎬ쪹컒뗄퇐랢릤ퟷ뗃틔쮳샻췪돉ꆣ ퟮ뫳ꎬ룐킻쯹폐뛔싛컄쳡돶놦맳틢볻뗄샏쪦뫍춬톧ꎬ룐킻퓚냙쎦횮훐움퓄싛컄뫍닎볓싛컄듰뇧뗄룷캻볒ꆣ ퟷ헟ꎺ 뚭좫뗂 2010쓪10퓂10죕 39
