Matrix embedding method in match for session-based recommendation

TL;DR

This paper introduces a matrix embedding approach for session-based recommendation systems, enhancing the representation of user interests by capturing interest diversity through quadratic forms, and demonstrates its superiority over traditional vector embeddings.

Contribution

The paper proposes a novel symmetric matrix embedding for sessions that models user interest diversity and improves matching accuracy in large-scale recommendation systems.

Findings

Matrix embedding outperforms vector embedding in experiments.

Eigenvectors of the matrix relate to user interests.

Method is effective for large-scale 'match' stage.

Abstract

Session based model is widely used in recommend system. It use the user click sequence as input of a Recurrent Neural Network (RNN), and get the output of the RNN network as the vector embedding of the session, and use the inner product of the vector embedding of session and the vector embedding of the next item as the score that is the metric of the interest to the next item. This method can be used for the "match" stage for the recommendation system whose item number is very big by using some index method like KD-Tree or Ball-Tree and etc.. But this method repudiate the variousness of the interest of user in a session. We generated the model to modify the vector embedding of session to a symmetric matrix embedding, that is equivalent to a quadratic form on the vector space of items. The score is builded as the value of the vector embedding of next item under the quadratic form. The eigenvectors of the symmetric matrix embedding corresponding to the positive eigenvalues are conjectured to represent the interests of user in the session. This method can be used for the "match" stage also. The experiments show that this method is better than the method of vector embedding.