Content Recommendation System Based on Private Dynamic User Profile

Internet is overwhelming, personalized content recommendation system offers spam filtering service and suggests useful information to the end users. It is a hotspot in the research area of content management on WWW. Traditional recommendation systems do the data mining on web access logs, discover user's access patterns, and filter the information on behalf of the user at the server side. One critical limitation of traditional recommendation system is the lack of user's private daily data, such as schedules, favorite websites and personal emails. The reason for this limitation is the privacy leak issue when the server holds much more private user data. To solve this problem, this paper presents an agent-based personalized recommendation method called Content REcommendation System based on private Dynamic User Profile (CRESDUP). The system collects and mines the private data of user at the client side, discovers, stores and updates private Dynamic User Profile (DUP) at the client side. The system fetches preferred message from the content server according to DUP. An important usage of this technology is a personalized advertising system in the RSS (Rich Site Summary, or RDF Site Summary) reader application. Our experiment shows that the system can utilize DUP to identify the customers' potential preferences and deliver the more preferred messages, especially the advertisements, to people who are interested.

[1]  MAGDALINI EIRINAKI,et al.  Web mining for web personalization , 2003, TOIT.

[2]  Michael J. Pazzani,et al.  Learning and Revising User Profiles: The Identification of Interesting Web Sites , 1997, Machine Learning.

[3]  Spiros Sirmakessis,et al.  Mining for Web Personalization , 2008 .

[4]  Joshua Goodman,et al.  Stopping outgoing spam , 2004, EC '04.

[5]  Ron Kohavi,et al.  Mining e-commerce data: the good, the bad, and the ugly , 2001, KDD '01.

[6]  Fabrizio Silvestri,et al.  Dynamic personalization of web sites without user intervention , 2007, CACM.

[7]  Nicholas J. Belkin,et al.  Information filtering and information retrieval: two sides of the same coin? , 1992, CACM.

[8]  Mark Rosenstein,et al.  Recommending and evaluating choices in a virtual community of use , 1995, CHI '95.

[9]  Bradley N. Miller,et al.  MovieLens unplugged: experiences with an occasionally connected recommender system , 2003, IUI '03.

[10]  Susan T. Dumais,et al.  A Bayesian Approach to Filtering Junk E-Mail , 1998, AAAI 1998.

[11]  Rick Wash,et al.  An economic answer to unsolicited communication , 2004, EC '04.

[12]  Li-Ping Jing,et al.  Improved feature selection approach TFIDF in text mining , 2002, Proceedings. International Conference on Machine Learning and Cybernetics.

[13]  Flavia Sparacino.  Sto(ry)chastics: A Bayesian Network Architecture for User Modeling and Compuatational Storytelling for Interactive Spaces , 2003, UbiComp.

[14]  CroftW. Bruce,et al.  Information filtering and information retrieval , 1992 .

[15]  Raymond J. Mooney and Paul N. Bennett and Loriene Roy,et al.  Book Recommending Using Text Categorization with Extracted Information , 1998 .

[16]  Ali F. Farhoomand,et al.  Managerial information overload , 2002, CACM.

[17]  Gordon V. Cormack,et al.  Spam and the ongoing battle for the inbox , 2007, CACM.

[18]  Yoav Shoham,et al.  Fab: content-based, collaborative recommendation , 1997, CACM.