摘要
随着Internet和信息技术的飞速发展,电子商务个性化推荐作为一种崭新的智能信息服务方式,针对不同的用户,通过对用户个性、习惯、偏好的分析,准确地向用户提供感兴趣的信息和服务,满足用户对于个性化产品需求的同时,提高了企业的竞争能力,得到了广泛重视。
在大众消费能力逐渐提高、选择产品变得更为精细的今天,汽车作为日常生活中越来越普及的交通工具,人们对其个性化的需求越来越大。由于汽车配置属性参数繁多,其用户群体更需要个性化推荐服务帮助其进行购买决策,个性化推荐系统成为解决这一问题的有效工具。
本文在借鉴国内外研究成果的基础上,通过分析汽车电子商务的特点,对协同过滤推荐算法及其在汽车电子商务中的应用进行了研究,主要工作包括:
(1)分别从汽车销售商和用户的角度,分析汽车电子商务个性化推荐服务的重要性,并对电子商务个性化推荐系统的组成结构和整体框架,以及主要的推荐技术进行了比较、分析。
(2)针对传统协同过滤推荐算法评分数据稀疏性和冷开始问题,采用基于属性值偏好矩阵的协同过滤算法,通过对矩阵的降维以及构建属性值偏好矩阵进行相似性度量,有效缓解稀疏性;同时建立用户反馈机制,降低冷启动问题的发生概率。
(3)将上述协同过滤算法应用到汽车电子商务的个性化推荐服务中,构造了汽车电子商务推荐系统模型,设计开发了原型系统,提供了实际应用的基础。
With the rapid development of Internet and information technology, e-commerce personalized recommendation as a kind of new way of intelligent information services provides the information and services that the users interested in exactly for different users through analyzing the users' personality, habits and preferences. It satisfies the users' demand for personalized products greatly and improves the competitiveness of enterprises.
Nowadays, the public consumption ability gradually increases and their products chosen become more and more sophisticated. Meanwhile, car, as the most popular vehicle in daily life, people's individual demands are growing towards it. Due to the parameter of the car configuration attributes are various, the consumer especially need personalized recommendation services to help them make purchasing decision. Therefore, the recommendation services system is the most effective tool to solve the problem.
Based on the research accomplishment both domestic and overseas and analyzing the characteristics of automotive e-commerce, the paper studied on the collaborative filtering algorithm and its application in automotive e-commerce, mainly include:
1) The paper analyzed the necessity of automotive e-commerce personalized recommendation service from the vendor and users, compared and analyzed the architecture, the overall framework of e-commerce personalized recommendation system and the main recommendation technology.
(2) In accordance with the problems of score sparsity and cold-start of traditional collaborative filtering recommendation algorithm, the paper adopted the collaborative filtering algorithm based on attribute value preference matrix. Through matrix dimensionality and building the attribute value preference matrix to measure similarity, sparsity has been alleviated effectively; while establishing user feedback mechanism reduce cold start.
(3) The paper put the collaborative filtering algorithm into the automotive e-commerce personalized recommendation services, and build a model of automotive e-commerce recommendation, designed and developed a prototype system. It provides a basis for practical application.
引文
[1]Ben S J, Konstan J A, Riedl J. E—commerce Recommendation Applications[EB/OL]. http: //www.grouplens.org/papers/pdf/ECRA.pdf,2002-09-26.
[2]Efraim Turban等著.Electronic commerce—A Managerial Perspective[M].高等教育出版社,317—329(2001)
[3]杨乔林.Internet上的电子商务[J],计算机系统应用,11:10-12(1998)
[4]United Nations E-commerce Develop Report[M].United Nations Publication,52 (2004)
[5]Maltz D, Miller B, Konstan J. Applying collabrative filtering to usenet new. Communication of the ACM,1997,40(3):77-87
[6]Maes P, Shardanand U. Algorithms for automating word of mouth. Proceedings of the ACM CHI'95 Conference on Human Factors in Computing Systems. New York:ACM Press,1995: 210-217
[7]Stead L, Hill W, Rosenstein M, Furnas G Recommending and evaluating choices in a virtual community of use. In CHI'95:Conference Proceedings on Human Factors in Computing Systems, Denver: CO,1995:194-201
[8]Kun-Lung W, Charu C A, Joel L W. A new graph-theoretic approach to collaborative filtering. In Proceedings of ACM SIGKDD International Conference on Knowledge Discovery &Data Mining,1999:45-63
[9]Heckerman D, Breese J, et al. Empirical analysis of predictive algorithm for collaborative filtering. In Proceedings of the 14 Conference on Uncertainly in Artificial Intelligence,1998: 43-52
[10]Pazzani M J. A framework for collaborative content-based and demographic filtering Artificial Review[J],1999,13(5/6):393-408
[11]Moukas A, Guttman R H. Agent-Mediated Electronic Commerce: A Survey. Knowledge Engineering Reviw,1998,13(2):147-159
[12]Bertram C, Sheth A, et al. Managing Semantic Content for the Web[M],2002(4):80-87
[13]Shoham Y, Balabanovic M. Content-Based Collaborative Recommendation. Communication of the ACM,1997(3):66-72
[14]Aggarwal C, Wu K L, et al. Personalization with dynamic profiler. In:Proceedings of the 3rd International Workshop on advanced Issues of E-Commerce,2000
[15]何军,周明天.信息网络中的信息过滤技术[J].系统工程与电子技术2001(n):76—79
[16]赵立江,何钦铭.一种个性化Web推荐系统的研究与实现[J].武汉理工人学学报,2004(5):681—684
[17]陈小华,赵捧未.基于关联规则的个性化信息检索系统研究[J].情报科学,2006(6):915—918
[18]冯翱,刘斌,卢增祥等.基于Agent的信息过滤系统[J].清华大学学报(自然科学版),2001(3):85—55
[19]WANG Weiping, Wu Lun. Application of Collaborative Filtering in CRM-Based Cross-Selling[J]. Journal of Management,2007,4(4):436-441.
[20]Byeong Man Kim, Qing Li,Jong-Wan Kim.A New collaborative recommender system addressing three Problerns.Proeeedings of PRICAI 2004,2004:495 — 504
[21]Massa P and Avesani P.Trust-aware Collaborative Filtering for Reeommender Systems. Proeeedings of Iniernational Conference on Cooperative Information Systems 2004,2004:120-160
[22]Mobasher B, Dai H, Luo T, et al.Diseovery of Aggregate Usage profiles for web Personalization.ACM WebKDD 2000 Web Mining for E-Commerce Workshop,2000
[23]Paul Resnick, Hal R. Varian, Recommender systems, Communications of the ACM [],1997.40(3)56-58;
[24]J.Ben Schafer, Joseph Konstan, John Riedl, Recommender Systems in E-Commerce,In ACM Conference on Electronic Commerce(EC99),1999;
[25]SchaferJ. B, KonstanLA, RiedlJ. Reeommender Systems in E-Commerce. In ACM ConfereneeonEleetronicCommeree(EC99),1999.
[26]G. Adomavicius and A. Tuzhilin. Expert-driven validation of rule-based user models in Personalizalion application.2000
[27]Lang, K.:1995, Newsweeder:Learning to flilter news In:Proeeedings of the 12th Intenational Conference on Machine Learning, LakeTahoe, CA, PP.331 — 339.
[28]Pazzani, M. J. A framework for collaborative, content-based an Demographic flitering.Artificial Intelligence Review.1999,13(5 — 6):393-408.
[29]林若萍.以从众化机制过滤超载信息之效果研究[D].台湾国立中正大学资讯管理研究所,2002
[30]Schafer, J. B., Konstan, J. A and Riedl, J.E-Commerce Recommendation Applications. Data Mining and Knowledge Discovery,2001
[31]黎星星,黄小琴,朱庆生.《电子商务推荐系统研究》[J], 《计算机工程与科学》,2004年,第26卷第5期
[32]Sarwar B, Karypis G, KonstanJ, Riedl J. Item-Based collaborative filtering recommendational gorithms. In:Proeeedings of the 10th International World Wide Web Conference.2001:285 — 295.
[33]George Karypis. Evaluation if Item-based Top-N Recommendation Algorithms. www. cs. umn. edu
[34]邓爱林,左子叶,朱扬勇.基于项目聚类的协同过滤推荐算法[J].小型微型计算机系统,2004,25(9):1665-1670.
[35]孙小华.协同过滤系统的稀疏性与冷启动问题研究[D].杭州:浙江大学,2005.
[36]Derwester S, Dumais ST, FumasGW, etal. Indexing by Latent Semantic Analysis [J]. Journal of the American Society for Information Science,1990,41(6):391-407.
[37]Sarwar BM, KarypisG, Konstan JA, etal. Application of Dimensionality Reduction in Recommender System-a case study[R]. Minneapolis, MN:University of Minnesota, 2000.
[38]余力.电子商务个性化—理论、方法与应用[M],清华大学出版社,2007.1
[39]Sarwar B, Karypis G, KonstanJ, Riedl J. Item-Based collaborative filtering recommendational gorithms. In:Proeeedings of the 10th International World Wide Web Conference.2001:285—295.
[40]黄巧莉,刘胜,刘飞.网络化销售和定制个性化信息推荐系统研究及应用[J].现代制造工程,2005,(8):31-35.
[41]曾子明,余小鹏.电子商务推荐系统与智能谈判技术[M],武汉大学出版社,2008.