Personalized recommendation system based on product specification values

摘要

In this paper, we developed a recommendation system which enables bidirectional communication between the user and system using an utility range-based product recommendation algorithm in order to provide more dynamic and personalized recommendations. The system is based on an interactive procedure for recommending similar ones among the products of the collaborative companies that share the product taxonomy table. The main idea of the proposed procedure is using a multi-attribute decision making (MADM) to find the utility values of products in same product class of the companies. Based on the values, we determine what products are similar. The similar product recommendation system is a Web-based application system running on a PC. The system has a user-friendly graphic user interface to encode easily incomplete value judgments. Using the system, we carry out the experiments for performance evaluation of our procedure. The experimental study shows that the utility range-based approach is a viable solution to the similar product recommendation problems in the viewpoints of correct rate and satisfaction rate.