Item Recommendation by Query-Based Biclustering Method

We propose a new recommender system that explores useful items by a biclustering method based on user's query. The advantage of our method is that the computational time can be reduced because the search space of biclusters is restricted to the transactions (users) which rate items within a query. In this study, the performance of our method is compared to that of a previous method that executes biclustering for entire transaction database. As a result, it is shown that our method enables item recommendation with higher accuracy at a considerably lower computational cost than the previous method.