Advancements in Sequential Recommendation Systems

The field of sequential recommendation systems is advancing rapidly, with a focus on improving the accuracy and transparency of recommendations. Researchers are exploring new approaches to leverage system exposure data, such as counterfactual augmentation and reinforcement learning, to better model user behavior and preferences. Another key direction is the development of more effective and efficient models for session-based recommendation, including the integration of linear and neural knowledge. Additionally, there is a growing interest in explainability and transparency, with methods being proposed to provide factual and counterfactual explanations for recommendations. Noteworthy papers in this area include:

  • A paper proposing a novel framework for factual and counterfactual explanations in session-based recommendation, which leverages reinforcement learning to uncover the minimal yet critical sequence of items influencing recommendations.
  • A paper introducing a reinforcement learning-based approach to address popularity bias and the cold-start problem in third-party library recommendations, which utilizes a graph convolution network-based embedding model and a carefully designed reward function.
  • A paper proposing a novel session-based recommendation model that integrates linear and neural knowledge, achieving significant improvements in accuracy and inference speed.

Sources

Improving Sequential Recommenders through Counterfactual Augmentation of System Exposure

Adaptive Long-term Embedding with Denoising and Augmentation for Recommendation

A Reinforcement Learning Method to Factual and Counterfactual Explanations for Session-based Recommendation

Bake Two Cakes with One Oven: RL for Defusing Popularity Bias and Cold-start in Third-Party Library Recommendations

Linear Item-Item Model with Neural Knowledge for Session-based Recommendation

Intent-aware Diffusion with Contrastive Learning for Sequential Recommendation

Built with on top of