Intent Disentanglement and Feature Self-supervision for Novel Recommendation

TL;DR

This paper introduces a novel recommendation framework that disentangles user intent into popularity and preference, improving novelty and accuracy, especially for cold-start tail items, through self-supervised learning and end-to-end optimization.

Contribution

It proposes a unified model that separates user intent into conformity and personal interest, and employs self-supervised learning for cold-start items, advancing the state-of-the-art in novelty and accuracy.

Findings

Significant improvements in accuracy and novelty over baselines.

Effective modeling of cold-start tail items using self-supervised learning.

Enhanced coverage and trade-off between accuracy and novelty.

Abstract

One key property in recommender systems is the long-tail distribution in user-item interactions where most items only have few user feedback. Improving the recommendation of tail items can promote novelty and bring positive effects to both users and providers, and thus is a desirable property of recommender systems. Current novel recommendation studies over-emphasize the importance of tail items without differentiating the degree of users' intent on popularity and often incur a sharp decline of accuracy. Moreover, none of existing methods has ever taken the extreme case of tail items, i.e., cold-start items without any interaction, into consideration. In this work, we first disclose the mechanism that drives a user's interaction towards popular or niche items by disentangling her intent into conformity influence (popularity) and personal interests (preference). We then present a unified end-to-end framework to simultaneously optimize accuracy and novelty targets based on the disentangled intent of popularity and that of preference. We further develop a new paradigm for novel recommendation of cold-start items which exploits the self-supervised learning technique to model the correlation between collaborative features and content features. We conduct extensive experimental results on three real-world datasets. The results demonstrate that our proposed model yields significant improvements over the state-of-the-art baselines in terms of accuracy, novelty, coverage, and trade-off.