資源描述:
《基于協(xié)同過(guò)濾算法的交互式問(wèn)答系統(tǒng)分析》由會(huì)員上傳分享,免費(fèi)在線(xiàn)閱讀,更多相關(guān)內(nèi)容在學(xué)術(shù)論文-天天文庫(kù)。
1、摘要摘要問(wèn)答系統(tǒng)可以被分為自動(dòng)問(wèn)答系統(tǒng)和交互式問(wèn)答系統(tǒng)。在電子商務(wù)等系統(tǒng)中,常見(jiàn)的問(wèn)答系統(tǒng)有客戶(hù)服務(wù)系統(tǒng),CallCenter,智能查詢(xún)等。但是目前的這些問(wèn)答系統(tǒng)中,并沒(méi)有根據(jù)問(wèn)題的有效期、用戶(hù)的知識(shí)結(jié)構(gòu)及積極性去設(shè)計(jì)個(gè)性化的推薦方案。協(xié)同過(guò)濾是現(xiàn)階段最成功的個(gè)性化推薦技術(shù)之一,怎樣將協(xié)同過(guò)濾算法應(yīng)用到交互式問(wèn)答系統(tǒng)中,是一項(xiàng)值得研究和解決的問(wèn)題。本文根據(jù)交互式問(wèn)答領(lǐng)域以及協(xié)同過(guò)濾算法的特點(diǎn),提出了基于協(xié)同過(guò)濾算法的交互式問(wèn)答系統(tǒng),對(duì)更加合理和有效地向用戶(hù)提供個(gè)性化推薦的相關(guān)問(wèn)題方面進(jìn)行了探索,以便使用戶(hù)準(zhǔn)確地得到
2、感興趣的問(wèn)題。本文的研究?jī)?nèi)容如下:第一,研究了交互式問(wèn)答系統(tǒng)的幾種常見(jiàn)模型,語(yǔ)義模式以及基于案例推理模型,并對(duì)基于案例推理模型進(jìn)行了改進(jìn)。第二,提出了基于協(xié)同過(guò)濾算法的交互式問(wèn)答系統(tǒng)。傳統(tǒng)的交互式問(wèn)答系統(tǒng)中,有的沒(méi)有提供任何問(wèn)題推薦機(jī)制,有的僅向用戶(hù)隨機(jī)推薦若干與他所回答問(wèn)題所屬版面相同的問(wèn)題,并沒(méi)有根據(jù)問(wèn)題的有效期、用戶(hù)的知識(shí)結(jié)構(gòu)及積極性去設(shè)計(jì)個(gè)性化的推薦方案.本文提出了基于協(xié)同過(guò)濾算法的交互式問(wèn)答系統(tǒng),從而向用戶(hù)提供更加合理和有效地個(gè)性化推薦。1、提出了基于用戶(hù)知識(shí)、用戶(hù)行為和用戶(hù)社會(huì)的用戶(hù)模型。2、改進(jìn)了基于
3、協(xié)同過(guò)濾算法的問(wèn)題推薦機(jī)制,通過(guò)計(jì)算評(píng)分矩陣中未評(píng)分問(wèn)題,在用戶(hù)瀏覽一個(gè)問(wèn)題的同時(shí),系統(tǒng)將選出對(duì)于該用戶(hù)來(lái)說(shuō)預(yù)測(cè)評(píng)分最高的若干個(gè)問(wèn)題,作為推薦問(wèn)題,顯示給用戶(hù)。第三,對(duì)基于協(xié)同過(guò)濾算法的交互式問(wèn)答系統(tǒng)進(jìn)行了分析和設(shè)計(jì),闡釋了該系統(tǒng)中的用戶(hù)數(shù)據(jù)庫(kù),用戶(hù)管理和建模模塊、問(wèn)題分析模塊和問(wèn)題推薦模塊,并介紹了各個(gè)模塊的功能。關(guān)鍵詞:交互式問(wèn)答系統(tǒng);協(xié)同過(guò)濾算法;個(gè)性化推薦廣東工業(yè)大學(xué)碩士學(xué)位論丈ABSTRACTQAsystemscanbeclassifiedintoautomaticQAsystemanduser—inte
4、ractiveQAsystem.Inthee—commercesystems,commonQAsystemsincludecustomerservicesystem,CallCenterandintelligentinquiryetc.However,thecurrentQAsystemsdonotprovidepersonalizedrecommendationaccordingtothevalidityoftheproblem,knowledgestructureofusersandpositiveness.C
5、ollaborativefilteringisoneofthemostsuccessfulpersonalizedrecommendationtechniques,SOhowcollaborativefilteringalgorithmisappliedtoaninteractiveQAsystemisaworthystudyandsolveproblems.AccordingtothespecialtyofinteractiveQAareaandcollaborativefilteringalgorithm,th
6、ispaperproposedaninteractiveQAsystembasedoncollaborativefilteringalgorithm,andamorereasonableandeffectivetoprovidepersonalizedrecommendationoftherelevantissuesareexploredinordertogetusersinterestedintheproblemaccurately.Thisarticleincludesthefollowing:First,we
7、investigatedseveralcommonmodelsoftheinteractiveQAsystem,includingthesemanticmodelsandCBRmodel,andtheCBRmodelisimproved.Second,weproposedaninteractiveQAsystembasedoncollaborativefilteringalgorithm.TraditionalinteractiveQAsystemhardlyprovidedanyrecommendationmec
8、hanism.someonlyrecommendedanumberofquestionsheansweredinthesamepagestotheuserwithoutconsideringthevalidityofquestion,theuser’Sknowledgestructureandincentivetodesignpersonalizedreco