發(fā)布時間：2018-05-24 02:50

  本文選題：個性化推薦 + 推薦多樣性�。� 參考：《北京郵電大學(xué)》2017年博士論文

【摘要】：隨著Internet的迅猛發(fā)展,互聯(lián)網(wǎng)在人們生活中所扮演的角色不再只是連通不同地域、不同人群的工具,網(wǎng)絡(luò)已經(jīng)成為人們獲取信息、購物消費的不可或缺的方式。尤其是進(jìn)入21世紀(jì)后,電子商務(wù)在全世界范圍受到了高度的重視,這種買賣雙方不謀面、通過互聯(lián)網(wǎng)進(jìn)行交易的模式在信息技術(shù)的帶動下迅速崛起。然而,隨著網(wǎng)絡(luò)信息資源的規(guī)模不斷擴(kuò)大,消費者面對電子商務(wù)網(wǎng)站提供的海量信息時,很難做到在短時間內(nèi)瀏覽所有的產(chǎn)品,最終將導(dǎo)致信息過載,因此個性化推薦系統(tǒng)應(yīng)運(yùn)而生。個性化推薦對緩解信息過載、提升網(wǎng)絡(luò)購物效率、促進(jìn)產(chǎn)品銷售具有重要影響。然而,國內(nèi)外電子商務(wù)網(wǎng)站雖然采用了多種個性化的推薦形式來改進(jìn)用戶體驗,但是不恰當(dāng)?shù)耐扑]使消費者依然面臨信息過載的困境。如何提升個性化推薦的效果,受到了研究人員和電子商務(wù)網(wǎng)站管理者的廣泛關(guān)注。既有研究多數(shù)關(guān)注如何提高個性化推薦算法的精確度,或者關(guān)注如何降低消費者的感知風(fēng)險進(jìn)而提高滿意度。然而,鮮有研究基于消費者的決策過程來關(guān)注和探究個性化推薦過程中推薦的多樣性對推薦效果會產(chǎn)生如何的影響。針對先前研究中存在的不足,本文從消費者的決策過程出發(fā),基于消費者兩階段決策理論、偏好不一致悖論、長尾理論以及錨定效應(yīng)等理論,綜合運(yùn)用管理學(xué)、營銷學(xué)、計算機(jī)科學(xué)等學(xué)科的思想和方法,對電子商務(wù)網(wǎng)站中個性化推薦的多樣性對推薦效果的影響進(jìn)行了實證分析和建模研究。本文的實證分析結(jié)果與理論預(yù)測結(jié)果基本一致,具有一定的科研價值與實踐價值。本文的主要研究結(jié)論和研究成果如下:第一、建立了推薦時機(jī)與推薦產(chǎn)品組合對推薦效果影響的理論模型,研究了不同時機(jī)下向消費者推薦不同的產(chǎn)品組合對個性化推薦效果的影響。本文基于消費者兩階段決策理論與偏好不一致悖論相應(yīng)的將推薦時機(jī)劃分為兩個階段,即形成考慮集合和做出最終選擇前后兩個階段。此外,本文將所推薦產(chǎn)品的劃分為同類產(chǎn)品與相關(guān)產(chǎn)品來進(jìn)行研究,并結(jié)合推薦時機(jī)與產(chǎn)品組合兩個因素構(gòu)建了推薦時機(jī)與推薦產(chǎn)品組合對推薦效果影響的理論模型,研究了不同時機(jī)下向消費者推薦不同的產(chǎn)品組合對個性化推薦效果的影響。得到的結(jié)論有:(1)與在決策第一階段收到產(chǎn)品推薦相比,消費者更傾向于采納決策第二階段收到的產(chǎn)品推薦;(2)在考慮集合形成階段中,消費者的關(guān)注對象并沒有明顯集中于同類產(chǎn)品,而對推薦產(chǎn)品的多樣性的訴求更為突出;(3)在最終做出消費決策時,消費者開始關(guān)注相關(guān)產(chǎn)品的推薦,考慮目標(biāo)產(chǎn)品以外的購買。第二、建立了推薦產(chǎn)品銷量與推薦產(chǎn)品評分對推薦效果影響的理論模型,研究了推薦產(chǎn)品的銷量與評分兩個因素單獨作用下和交互作用下對推薦效果的影響。本文將推薦的產(chǎn)品劃分為主流產(chǎn)品與利基產(chǎn)品來進(jìn)行研究,并選取了產(chǎn)品評論的其中一種形式——產(chǎn)品的評分結(jié)合產(chǎn)品的銷量構(gòu)建了推薦產(chǎn)品的銷量與推薦產(chǎn)品評分對推薦效果影響的理論模型,分別探究了這兩個因素在不同的推薦情景中單獨作用或交互作用下對推薦效果產(chǎn)生的影響。得到的結(jié)論有:(1)與僅推薦主流產(chǎn)品或僅推薦利基產(chǎn)品相比,當(dāng)系統(tǒng)同時推薦主流與利基產(chǎn)品時,消費者會采納更多的推薦產(chǎn)品;(2)與僅推薦高評分產(chǎn)品或僅推薦低評分產(chǎn)品相比,當(dāng)系統(tǒng)同時推薦這兩種產(chǎn)品時,并沒有改善推薦效果,消費者并不會多的去購買系統(tǒng)所推薦的產(chǎn)品;(3)對比系統(tǒng)僅推薦高評分主流產(chǎn)品或僅推薦低評分利基產(chǎn)品的情況,當(dāng)系統(tǒng)同時推薦二者時,消費者更傾向于采納系統(tǒng)所推薦的產(chǎn)品;而對比系統(tǒng)僅推薦高評分利基產(chǎn)品或僅推薦低評分主流產(chǎn)品的情況,當(dāng)系統(tǒng)同時推薦二者時,消費者更傾向于采納系統(tǒng)所推薦的產(chǎn)品。第三、在構(gòu)建兩個理論模型的同時,相應(yīng)的將推薦產(chǎn)品劃分不同的類型,進(jìn)一步探索了消費者對不同類型的產(chǎn)品推薦采納的差異,推動了管理與消費者行為視角下的個性化推薦多樣性的研究。本文分別從營銷學(xué)與信息經(jīng)濟(jì)學(xué)兩個不同的學(xué)科視角,以公認(rèn)的產(chǎn)品分類方式對應(yīng)的將推薦的產(chǎn)品劃分為享樂品和實用品、搜索品和體驗品等多個不同的類型。在構(gòu)建推薦時機(jī)與產(chǎn)品組合對推薦效果影響的理論模型時將享樂品和實用品作為調(diào)節(jié)變量,同時在構(gòu)建推薦產(chǎn)品銷量與評分對推薦效果影響的理論模型時將搜索品和體驗品作為調(diào)節(jié)變量,進(jìn)一步探索了消費者對不同類型的產(chǎn)品個性化推薦的采納差異。研究結(jié)果顯示,消費者對不同類型的產(chǎn)品存在推薦采納的差異,享樂品的推薦效果好于實用品的推薦效果、體驗品的推薦效果好于搜索品的推薦效果。本文的理論貢獻(xiàn)主要體現(xiàn)在以下三個方面:第一,本文從消費者的決策過程出發(fā),揭示了消費者對個性化推薦的采納機(jī)制;第二,本文探索了個性化推薦的多樣性對推薦效果的影響,驗證了消費者對推薦內(nèi)容多樣性的需求;第三,本文進(jìn)一步探究了不同的產(chǎn)品分類對推薦效果產(chǎn)生的影響,揭示了消費者對不同類型產(chǎn)品個性化推薦的采納差異。本文對實踐的啟示意義包括以下兩個方面:第一,本文強(qiáng)調(diào)應(yīng)充分考慮消費者在不同決策階段的行為特點來改善個性化推薦系統(tǒng)的設(shè)計,在考慮推薦精確性的同時充分重視推薦內(nèi)容的多樣性;第二,本文強(qiáng)調(diào)應(yīng)充分利用消費者對享樂品和體驗品推薦內(nèi)容的強(qiáng)偏好性,通過個性化推薦來促進(jìn)這兩種產(chǎn)品的銷售,而對于實用品和搜索品,除了繼續(xù)完善推薦算法外,還需進(jìn)一步探索與其他營銷手段相結(jié)合來促進(jìn)產(chǎn)品的銷售。
[Abstract]:With the rapid development of Internet, the role of Internet in people's life is no longer only a way to connect different regions and different groups of tools. The network has become an indispensable way for people to obtain information and purchase consumption. Especially after entering twenty-first Century, e-commerce has been highly valued all over the world. However, with the increasing scale of the network information resources, it is difficult for consumers to browse all the products in a short time as the scale of the network information resources is expanding, and it will eventually lead to information overload, so personalized recommendation. The personalized recommendation has an important impact on alleviating the information overload, improving the network shopping efficiency and promoting the product sales. However, the domestic and foreign e-commerce websites have adopted a variety of personalized recommendation forms to improve the user experience, but the inappropriate recommendation makes the consumers still face the difficult situation of information overload. The effect of the promotion of personalized recommendation has attracted wide attention from researchers and E - commerce website managers. Most of the research focuses on how to improve the accuracy of personalized recommendation algorithms or how to reduce the perceived risk of consumers to improve their satisfaction. According to the shortcomings of the previous research, this paper, based on the consumer's decision-making process, based on the consumer's two stage decision theory, the preference inconsistency paradox, the long tail theory and the anchoring effect, makes a comprehensive use of management, marketing, and calculation. The influence of the diversity of personalized recommendation on the recommendation effect in e-commerce websites is analyzed and modeled. The results of this paper are basically consistent with the theoretical prediction results, and have certain scientific research value and practical value. The main research conclusions and research results of this paper are as follows. First, the theoretical model of the effect of recommendation time and recommended product combination on the effect of recommendation is set up, and the effect of recommending different product combinations to consumers at different times is studied. This paper divides the recommendation time into two stages based on the corresponding two stage decision theory of consumers and the inconsistent paradox of preference. In addition, this paper divides the recommended products into two stages. In addition, this paper divides the recommended products into similar products and related products, and constructs a theoretical model of the effect of recommendation timing and recommended product combination on recommended effects combined with two factors of recommendation timing and product combination, and studies the different opportunities at different times. The effect of different product combinations on Personalized Recommendation effect is recommended to consumers. The conclusions are as follows: (1) consumers are more inclined to adopt product recommendation in the second stage of decision making than to receive product recommendation in the first stage of decision making; (2) in consideration of the formation stage, the consumer's concern is not obviously concentrated in the same category. The demand for the diversity of recommended products is more prominent; (3) in the end of the decision making, consumers begin to pay attention to the recommendation of the related products and consider the purchase outside the target product. Second, a theoretical model of the effect of the recommended product sales and the recommended product score on the recommendation effect is established, and the sales and evaluation of the recommended products are studied. This paper divides the recommended products into the mainstream and niche products, and selects one of the forms of product reviews - the product score and the sales volume of the product to construct the recommended product sales and recommended product scores on the recommendation effect. Two The theoretical model of fruit impact examines the effects of these two factors separately or interacting on the different recommended scenarios. The conclusions are as follows: (1) consumers will adopt more recommendations when the system recommends the main stream and niche products at the same time compared with only the mainstream products or only the niche products recommended. Products; (2) when compared with only highly rated products or only recommended low grade products, when the system recommends these two products at the same time, it does not improve the recommendation effect. Consumers will not be able to buy more products recommended by the system; (3) the contrast system recommends only high grade mainstream products or only low score niche products, when the system is the same When the two are recommended, consumers are more inclined to adopt the products recommended by the system; and the contrast system recommends only high grade niche products or only the low score mainstream products. When the system recommends the two, the consumer is more inclined to adopt the products recommended by the system. Third, while constructing the two theoretical models, the corresponding will be Recommending different types of products, further exploring the differences between consumers and different types of products recommended by consumers, and promoting the diversity of personalized recommendation from the perspective of management and consumer behavior. This paper, from two different disciplinary perspectives of marketing and information economics, will correspond to recognized product classification methods. The recommended products are divided into hedonic and practical products, search products and experiential products, such as many different types. The results show that consumers have recommended differences in different types of products, and that the recommendation effect of pleasure products is better than that of practical products. The recommendation effect of experiential products is better than that of search products. The theoretical contribution of this article is mainly embodied in the following three aspects: first, this paper, starting from the decision-making process of the consumer, reveals the adoption mechanism of the consumer's personalized recommendation; second, this article explores the influence of the diversity of personalized recommendation on the recommendation effect, and validating the consumer's demand for the diversity of the recommended content; third, This paper further explores the impact of different product classifications on the effect of recommendation effect, and reveals the adoption differences of consumers' personalized recommendation for different types of products. The implications for practice include the following two aspects: first, this article emphasizes that the behavior characteristics of consumers at different decision-making stages should be fully considered to improve individualization. Recommending the design of the system to give full attention to the diversity of the recommended content while considering the accuracy of the recommendation. Second, this article emphasizes that it should make full use of the strong preference of the consumer for the content of the pleasure and experience, and promote the sales of the two products by personalized recommendation, and to the practical and search products, in addition to continuing to perfect the recommendation. The algorithm also needs further exploration and other marketing means to promote product sales.
【學(xué)位授予單位】：北京郵電大學(xué)
【學(xué)位級別】：博士
【學(xué)位授予年份】：2017
【分類號】：F724.6;F274

【相似文獻(xiàn)】

相關(guān)期刊論文 前10條

1 劉蓓琳;;基于用戶滿意度的電子商務(wù)個性化推薦評價研究[J];中國物流與采購;2012年14期

2 姜有輝;高琳琦;;電子商場中的個性化推薦研究與應(yīng)用[J];商場現(xiàn)代化;2006年33期

3 郭斌;董斯維;;電子商務(wù)網(wǎng)站發(fā)展新方法——個性化推薦[J];消費導(dǎo)刊;2006年10期

4 袁寧;;消費者對網(wǎng)絡(luò)推薦的抗拒研究[J];現(xiàn)代商貿(mào)工業(yè);2011年23期

5 陳紅亮;周少華;;移動商務(wù)中基于定位的個性化推薦研究[J];統(tǒng)計與決策;2008年21期

6 梁英;;電子商務(wù)個性化推薦技術(shù)研究[J];商場現(xiàn)代化;2007年26期

7 張建偉;;基于社交網(wǎng)絡(luò)數(shù)據(jù)的個性化推薦研究[J];商;2013年24期

8 陳龍飛;趙雪;;關(guān)于農(nóng)資產(chǎn)品個性化推薦的研究[J];電子世界;2013年24期

9 樊靜;;基于OBUIM的電子商務(wù)個性化推薦模型研究[J];現(xiàn)代商貿(mào)工業(yè);2009年19期

10 王茜;錢力;;大數(shù)據(jù)環(huán)境下電子商務(wù)個性化推薦服務(wù)發(fā)展動向探析[J];商業(yè)研究;2014年08期

相關(guān)會議論文 前4條

1 馮時;陽峰;王大玲;于戈;;基于虛擬觀點社群的用戶個性化推薦[A];第六屆全國信息檢索學(xué)術(shù)會議論文集[C];2010年

2 劉晉;王潛平;房坤;;基于agent技術(shù)和反饋機(jī)制的個性化推薦方法研究與設(shè)計[A];第二屆和諧人機(jī)環(huán)境聯(lián)合學(xué)術(shù)會議(HHME2006)——第2屆中國人機(jī)交互學(xué)術(shù)會議(CHCI'06)論文集[C];2006年

3 王鑫;黃忠義;;網(wǎng)絡(luò)資源中基于K-Means聚類的個性化推薦[A];2013年全國通信軟件學(xué)術(shù)會議論文集[C];2013年

4 曹雷;杜輝鋒;;基于Web挖掘的個性化推薦模型的研究[A];全國第十屆企業(yè)信息化與工業(yè)工程學(xué)術(shù)年會論文集[C];2006年

相關(guān)重要報紙文章 前2條

1 云蹤;雙模式完善精準(zhǔn)個性化推薦[N];人民郵電;2014年

2 本報記者 李娟;百分點 給電商配備“導(dǎo)購員”[N];中國經(jīng)營報;2011年

相關(guān)博士學(xué)位論文 前6條

1 張琳;電子商務(wù)網(wǎng)站個性化推薦的多樣性對推薦效果的影響研究[D];北京郵電大學(xué);2017年

2 呂苗;基于情境的商品個性化推薦方法研究[D];大連理工大學(xué);2015年

3 易明;基于Web挖掘的電子商務(wù)個性化推薦機(jī)理與方法研究[D];華中科技大學(xué);2006年

4 邢星;社交網(wǎng)絡(luò)個性化推薦方法研究[D];大連海事大學(xué);2013年

5 張宜浩;基于半監(jiān)督學(xué)習(xí)的個性化推薦研究[D];重慶大學(xué);2014年

6 張富國;基于信任的電子商務(wù)個性化推薦關(guān)鍵問題研究[D];江西財經(jīng)大學(xué);2009年

相關(guān)碩士學(xué)位論文 前10條

1 馮超;電子商務(wù)個性化推薦的用戶信息采納影響因素研究[D];華中師范大學(xué);2015年

2 尚剛;個性化推薦技術(shù)在圖書館服務(wù)中的應(yīng)用研究[D];南京財經(jīng)大學(xué);2015年

3 林舉;基于完全三部圖模型的個性化推薦技術(shù)研究[D];上海大學(xué);2015年

4 周瑩;基于證券理財產(chǎn)品用戶行為分析的個性化推薦研究[D];電子科技大學(xué);2014年

5 李祉瑩;基于語義的個性化推薦模型研究[D];北京郵電大學(xué);2015年

6 劉洋;數(shù)字學(xué)習(xí)資源個性化推薦技術(shù)研究與應(yīng)用[D];浙江工商大學(xué);2015年

7 陳玉兆;大數(shù)據(jù)下的個性化推薦研究與實現(xiàn)[D];西安電子科技大學(xué);2014年

8 盧云騁;播存環(huán)境中話題挖掘及其在個性化推薦中的應(yīng)用[D];東南大學(xué);2015年

9 劉佳;基于情景分析的網(wǎng)絡(luò)閱讀個性化推薦方法研究[D];東北大學(xué);2014年

10 路文;基于相似案例分析的電子商務(wù)個性化推薦方法研究[D];東北大學(xué);2014年

，

本文編號：1927394