ARTICLE N PRESS xpert Systems with Applications xxx(2011)xXX-XXX Contents lists available at SciVerse Science Direct Expert Systems with Applications ELSEVIER journalhomepagewww.elsevier.com/locate/eswa Hybrid personalized recommender system using centering-bunching based clustering algorithm Subhash K. Shinde, Uday Kulkarni ollege of Engineering, Navi Mumbai 400 614, India Department of Computer Science and E ng, SGGS Institute of Engineering and Technology, Nanded, India ARTICLE INFO ABSTRACT n recent years, there is overload of products information on world wide web. a personalized recommen- elaborative filtering dation is an enabling mechanism to overcome information overload occurred when shopping in an Inte Web personalized recommender system net marketplace. This paper proposes a novel centering-bunching based clustering(CBBC) algorithm hich is used for hybrid personalized recommender system(CBBCHPRS). The proposed system works n two phases. In the first phase, opinions from the users are collected in the form of user-item rating matrix. They are clustered offline using CBBC into predetermined number clusters and stored in a data base for future recommendation. In the second phase, the recommendations are generated online for active user using similarity measures by choosing the clusters with good quality rating. This helps get further effectiveness and quality of recommendations for the active users. The experimental results sing Iris dataset show that the proposed CBBC performs better than K-means and new K-medodis algo thms. The performance of CBBCHPRS is evaluated using Jester database available on website of Califor- d compared with ants recommender system(ARS). The results obtained at the proposed CBBCHPRS performs superiorly and alleviates problems such as cold-start. fir sparsity. 2011 Elsevier Ltd. All rights reserved. 1 Introduction are applied in many areas such as: web-browsing, information fill- tering, net-news or movie recommender and e-commerce. The cen- Many e-commerce Web sites offer numerous services, so a prod- tral element of all recommender systems is the user model that uct search could return an overwhelming set of options. Without contains knowledge about the individual preferences which deter- system support, filtering irrelevant products, comparing alterna- mine his or her behavior in a complex environment of web-based tives, and selecting the best option can be difficult or impossible. system. The WPRs are characterized by cross-fertilization of address this problem and suggest products(movies, music, jokes, ligence, knowledge representation, discovery and data web mining. books, news, web pages) that suit the user's needs(Adomavicius computational learning and intelligent and adaptive agents. the Tuzhilin, 2005 ) These systems add related information of items alternating information environment that is combined of variou to the information flowing towards the user, as opposed to remov- users, their needs and contexts of use as well as different system g information items. typically a recommender system compares platforms necessitates application of recommender systems. the the users profile to some reference characteristics, and seeks to pre- ever increasing importance of the e-commerce in the global econ- dict the rating that a user would give to an item they had not yet omy also increases the importance of the WPRS. They are developed considered Recommender systems use collaborative filtering ap- by different domains such as personal agents and adaptive hyper proaches or a combination of the collaborative and content based media. the personalized hypermedia application is defined as a filtering approaches, although content-based recommender sys- hypermedia system that adapts: the content, structure and or pre- tems do exist(Koren, Bell, Chris, 2009). The web personalized re sentation of the web objects to each individual user's modeL emendation systems(WPRS)are recently applied to provide The remainder of this paper is organized as follows. The section different type of customized information for their users. The wPrs 2 describes various types of input data that are used for the recom- mendation systems. The section 3 summarizes the different tech- niques of recommendation systems and their drawbacks. The ding author..Tel+9109221788066;fax:+9102227473196 proposed clustering based hybrid personalized recommender E-mail addresses: skshindeerediffmail com(S.K. Shinde), kulkuniuveyahoo. system is described in the section 4. The section 5 illustrates exper- imental setup of the proposed recommendation system. This 0957-4174/s- see front matter o 2011 Elsevier Ltd. All rights reserved doi:10.1016/eswa2011.08020 Please cite this article in press as: Shinde, S.K.& Kulkarni, U Hybrid personalized recommender system using centering-bunching based clustering algo- rithm. Expert Systems with Applications(2011). doi: 10.1016/jeswa2011.08.020Hybrid personalized recommender system using centering-bunching based clustering algorithm Subhash K. Shinde ⇑ , Uday Kulkarni Department of Computer Engineering, Bharati Vidyapeeth College of Engineering, Navi Mumbai 400 614, India Department of Computer Science and Engineering, SGGS Institute of Engineering and Technology, Nanded, India article info Keywords: Collaborative filtering Centering-bunching based clustering Web personalized recommender system abstract In recent years, there is overload of products information on world wide web. A personalized recommendation is an enabling mechanism to overcome information overload occurred when shopping in an Internet marketplace. This paper proposes a novel centering-bunching based clustering (CBBC) algorithm which is used for hybrid personalized recommender system (CBBCHPRS). The proposed system works in two phases. In the first phase, opinions from the users are collected in the form of user-item rating matrix. They are clustered offline using CBBC into predetermined number clusters and stored in a database for future recommendation. In the second phase, the recommendations are generated online for active user using similarity measures by choosing the clusters with good quality rating. This helps to get further effectiveness and quality of recommendations for the active users. The experimental results using Iris dataset show that the proposed CBBC performs better than K-means and new K-medodis algorithms. The performance of CBBCHPRS is evaluated using Jester database available on website of California University, Berkeley and compared with ants recommender system (ARS). The results obtained empirically demonstrate that the proposed CBBCHPRS performs superiorly and alleviates problems such as cold-start, first-rater and sparsity. 2011 Elsevier Ltd. All rights reserved. 1. Introduction Many e-commerce Web sites offer numerous services, so a product search could return an overwhelming set of options. Without system support, filtering irrelevant products, comparing alternatives, and selecting the best option can be difficult or impossible. The recommender systems are personalized applications that can address this problem and suggest products (movies, music, jokes, books, news, web pages) that suit the user’s needs (Adomavicius & Tuzhilin, 2005). These systems add related information of items to the information flowing towards the user, as opposed to removing information items. Typically, a recommender system compares the user’s profile to some reference characteristics, and seeks to predict the rating that a user would give to an item they had not yet considered. Recommender systems use collaborative filtering approaches or a combination of the collaborative and content based filtering approaches, although content-based recommender systems do exist (Koren, Bell, & Chris, 2009). The web personalized recommendation systems (WPRS) are recently applied to provide different type of customized information for their users. The WPRS are applied in many areas such as: web-browsing, information filtering, net-news or movie recommender and e-commerce. The central element of all recommender systems is the user model that contains knowledge about the individual preferences which determine his or her behavior in a complex environment of web-based system. The WPRS are characterized by cross-fertilization of various research fields such as: information retrieval, artificial intelligence, knowledge representation, discovery and data/web mining, computational learning and intelligent and adaptive agents. The alternating information environment that is combined of various users, their needs and contexts of use as well as different system platforms necessitates application of recommender systems. The ever increasing importance of the e-commerce in the global economy also increases the importance of the WPRS. They are developed by different domains such as personal agents and adaptive hypermedia. The personalized hypermedia application is defined as a hypermedia system that adapts: the content, structure, and/or presentation of the web objects to each individual user’s model. The remainder of this paper is organized as follows. The section 2 describes various types of input data that are used for the recommendation systems. The section 3 summarizes the different techniques of recommendation systems and their drawbacks. The proposed clustering based hybrid personalized recommender system is described in the section 4. The section 5 illustrates experimental setup of the proposed recommendation system. This 0957-4174/$ - see front matter 2011 Elsevier Ltd. All rights reserved. doi:10.1016/j.eswa.2011.08.020 ⇑ Corresponding author. Tel.: +91 09221788066; fax: +91 022 27473196. E-mail addresses: skshinde@rediffmail.com (S.K. Shinde), kulkurniuv@yahoo. com (U. Kulkarni). Expert Systems with Applications xxx (2011) xxx–xxx Contents lists available at SciVerse ScienceDirect Expert Systems with Applications journal homepage: www.elsevier.com/locate/eswa Please cite this article in press as: Shinde, S. K., & Kulkarni, U. Hybrid personalized recommender system using centering-bunching based clustering algorithm. Expert Systems with Applications (2011), doi:10.1016/j.eswa.2011.08.020