Exploiting Session Information in BERT-based Session-aware Sequential Recommendation

Abstract

In recommendation systems, utilizing the user interaction history as sequential information has resulted in great performance improvement. However, in many online services, user interactions are commonly grouped by sessions that presumably share preferences, which requires a different approach from ordinary sequence representation techniques. To this end, sequence representation models with a hierarchical structure or various viewpoints have been developed but with a rather complex network structure. In this paper, we propose three methods to improve recommendation performance by exploiting session information while minimizing additional parameters in a BERT-based sequential recommendation model: using session tokens, adding session segment embeddings, and a time-aware self-attention. We demonstrate the feasibility of the proposed methods through experiments on widely used recommendation datasets.

6 pages, accepted in The 45th International ACM SIGIR Conference on Research and Development in Information Retrieval (SIGIR) 2022, short paper

References in corpus (10)

• Personalizing Session-based Recommendations with Hierarchical Recurrent Neural Networks
• Lightweight Self-Attentive Sequential Recommendation
• Sequential Recommendation with Self-Attentive Multi-Adversarial Network
• Session-aware Information Embedding for E-commerce Product Recommendation
• MEANTIME: Mixture of Attention Mechanisms with Multi-temporal Embeddings for Sequential Recommendation
• Exploiting Positional Information for Session-based Recommendation
• Unsupervised Proxy Selection for Session-based Recommender Systems
• Balancing Multi-level Interactions for Session-based Recommendation
• Regularized Adversarial Sampling and Deep Time-aware Attention for Click-Through Rate Prediction
• Rethinking Lifelong Sequential Recommendation with Incremental Multi-Interest Attention