Combining Interpretability and Filtering Algorithm for Sequence Recommendation

Abstract

Sequence recommendation has become more and more important in various online recommendation services. It can obtain the user's dynamic preference characteristics from the user's historical behavior data to predict the next possible interaction item of the user, so as to provide accurate recommendations for the user. Most of the existing models only focus on a single user history interaction sequence, which will lead to the lack of interpretability of the recommendation results. Meanwhile, in the online platform, the recorded user behavior data will inevitably contain noise. In order to solve the above problems, we propose a new method Combining Interpretability and Filtering algorithm for Sequence recommendation (CIFA_Rec). Through multi-attribute modeling, the interpretability of the sequence recommendation model is increased. At the same time, we also borrow the idea of filtering algorithm and use the learning filter to adaptively reduce the data noise. A large number of experiments on three real datasets show that our proposed model has good performance in sequence recommendation and interpretability.