Research On Dual Attentive Neural Network For Personalizing Session-based Recommendation

In recent years,Session-based Recommendation has been widely studied in recommendation system tasks.Session-based Recommendation aims to recommend the next item in an anonymous session for users.Traditional methods such as matrix factorization and item-to-item perform very poorly because they only take into account the last click of the session and ignore the information of the whole click sequence.Recurrent Neural Network(RNN)based methods have performed excellently in session-based recommendation,but they only consider the user's sequential behavior in the current session or only use crosssession information to track user's interests over time,so these works can not get a good recommendation result.In order to get better personalized recommendations for users and aim at the deficiency of current session-based recommendation models and algorithms.we design a novel neural network framework for personalized session-based recommendation,named Dual Attentive Neural Network(DANN).DANN considers user's main purpose of current session and user's personalized preference of cross-session.Specifically,in DANN we exploit a userlevel attention mechanism to model user's personalized preference and capture user's main purpose in the current session via a sessionlevel attention mechanism.And then we get a new fusion representation via encoder-decoder structure computes the hidden representation of user's purpose and the hidden representation of user's preference.Finally,the decoder decodes the new fusion representation and gets the final recommendation results list.Experiments on two real-world website data sets XING and Retailrocket show that our DANN model outperforms other baseline models in the two evaluation indexes recall value and MRR value.At the same time,the validity and rationality of our DANN model proposed in this paper are verified by a large number of experiments.