A Client/Server User-Based Collaborative Filtering Algorithm: Model and Implementation

Abstract. This paper describes a new way of implementing an intelligent web caching service, based on an analysis of usage. Since the cache size in software is limited, and the search for new information is time-consuming, it becomes interesting to automate the process of selecting the most relevant items for each user. We propose a new generic model based on a client/server collaborative filtering algorithm and a behavior modeling process. In order to highlight the benefits of our solution, we collaborated with a company called ASTRA which is specialized in satellite website broadcasting. ASTRA has finalized a system sponsored by advertisement and supplying to users a high bandwidth access to hundreds of websites for free. Our work has been implemented within its software architecture and, in particular, within its recommender system in order to improve the satisfaction of users. Our solution is particularly designed to address the issues of data sparsity, privacy and scalability. Bec...

• Type: Conference
• Year: 2006
• Where: ECAI
• Authors: Sylvain Castagnos, Anne Boyer
• Keywords: Artificial Intelligence, Collaborative Filtering Algorithm, Company Called Astra, ECAI 2006, Web Caching Service