An Effective Recommendation System for Querying Web Pages

In this paper, a recommendation system for querying web pages is developed. When users query web pages through a search engine, the query keywords, browsed web pages, and feedback values are collected as user query transactions. A clustering algorithm based on bipartite graph analysis is designed to determine clusters of query keywords and the browsed web pages, called access preference clusters. Next, association rules of query keywords and web pages are mined for each access preference cluster. The feedback values of browsed web pages are incorporated into the calculation of the support and confidence for each association rule to reflect the subjective opinions. Based on the mined clusters and rules, the system applies the concept of collaborative filtering to recommend highly semantics relevant web pages in the access preference clusters partially covered by a user profile or a given query. The initial experiment result shows the system can improve the querying effect of searching engines. * This work was partially supported by the Republic of China National Science Council under Contract No. 92-2213-E-003-012 Author to whom all correspondence should be addressed.