Processing Recommender Top-N Queries in Relational Databases

Abstract

According to the feedback information from a user in the result sets of initial or previous queries, we present in this paper a framework for processing recommender top-N queries in relational databases. Based on the techniques and ranking strategies of keyword search, this framework returns top-N results for an initial query given by the user. As soon as he or she selects some of the top-N results, the framework will find out related keywords from the result(s) selected by the user, calculate and modify corresponding weights of the related keywords. By using the weights, our framework determines new query words associated with the previous query to construct a recommender query. A knowledge base is created to store the related information of the tuples in the underlying database for evaluating the recommender query. The experimental results based on real datasets show the efficiency and effectiveness of our framework.