A User Preference Classification Method in Information Recommendation System

As information overload problem more serious on the Internet, it becomes an important issue for users to retrieve information effectively. An information recommendation system is helpful for providing user information meet he/she requirements appropriately. However the traditional recommendation concepts usual classify a user into one preference class. It seems unreasonable because a user may possess interests about many information classes generally. In this study we propose a new recommendation concept in information recommendation system, namely club member, differs from content-based and collaboration filter method. It can classify a user into some clubs which he/she interests with different preference degrees. In order to classify users into multi-club with different preference degrees, fuzzy association rule based on data mining technology is applied in this study. Fuzzy association rules are generated by discovering and analyzing the members’ feature in the same club. According to fuzzy association rules, a new user can be classified into some clubs that he/she may interest with preference degrees. It is helpful for an information recommendation system to provide user more suitable information in accordance with their preferences precisely