Host: The Japanese Society for Artificial Intelligence
Name : 34th Annual Conference, 2020
Number : 34
Location : Online
Date : June 09, 2020 - June 12, 2020
Recently, it has become possible to make use of not only simple purchase history but also acquire various types of auxiliary information. Therefore, it is expected to analyze these various kinds of data in order to investigate customers' purchasing behavior for marketing purposes. For that reason, the KGAT model, which learns user preferences by modeling the relationship between users, purchase items and their auxiliary information, has been proposed. In this model, the user's preference can be interpreted by using the auxiliary information of items and this interpretability can be useful for planning marketing policies. Therefore, this research proposes a model that enables more diversified analysis by extending KGAT by using not only auxiliary information of items but also the relationship between users and their attribute information. Finally, we apply the proposed method to the evaluation of historical data of actual EC sites and show the usefulness of the proposed method.