ファジィAHPを用いた製品レコメンドシステムの開発

抄録

Recently, recommendation systems and user reviews have attracted attention as aids in searching for desired products by using web information on the products. However, existing recommendation systems require analyzing a user's activity on the web and collecting information from many reviews. Therefore, in this study, we aimed to develop a product recommendation system that does not have these requirements. We used digital cameras and earphones as case studies and, on the basis of perception-of-value information obtained from questionnaires to users, we applied an analytic hierarchy process (AHP) to an analysis and calculated the degree-of-importance values of the evaluation items for each product. We then calculated the product evaluation values of each evaluation item by using user evaluations of products acquired from Kakaku.com Application Program Interface (API). We developed a system that decides the recommendation order of each product by multiplying the degree-of-importance values and evaluation values together by Fuzzy AHP.