PD-SRS: Personalized Diversity for a Fair Session-Based Recommendation System

Publication Name

Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)


Session-based Recommender Systems (SRSs), which aim to recommend users’ next action based on their current and historical sessions, play a significant role in many real-world online services. The existing session-based recommendation methods have mainly focused on the accuracy of recommendation, which biases to reinforce popular items/services and loses the recommendation diversity. Diversity is a positive aspect particularly in SRSs as the target user may like to be surprised and interact with a broader range of content in different sessions. In this work, we propose a Personalized Diversification strategy for a Session-based Recommender System (PD-SRS) using graph neural networks. Comprehensive experiments are carried out on two real-world datasets to demonstrate the effectiveness of PD-SRS in making a trade-off between accuracy and personalized diversity over the baselines.

Open Access Status

This publication is not available as open access


13740 LNCS

First Page


Last Page




Link to publisher version (DOI)