CROWN: A Novel Approach to Comprehending Users' Preferences for Accurate Personalized News Recommendation

2023-10-13

AI Summary

Personalized news recommendation faces three key challenges: modeling users’ multiple reading intents, distinguishing post-reading preferences, and handling cold-start users. To address these, we propose CROWN, an end-to-end framework featuring (1) a novel category-guided intent disentanglement mechanism that explicitly separates news items’ heterogeneous semantic intents; (2) consistency-driven news representation learning and GNN-enhanced hybrid user representation, jointly capturing content-level consistency and high-order user–news–category interactions; and (3) category prediction as an auxiliary task to strengthen supervision. By integrating intent disentanglement learning, consistency-aware contrastive learning, and multi-task joint optimization, CROWN achieves significant improvements over ten state-of-the-art methods on two real-world datasets. Ablation studies confirm the effectiveness of each component. CROWN establishes a new paradigm for multi-intent news recommendation—interpretable, robust, and generalizable.

Abstract

Personalized news recommendation aims to assist users in finding news articles that align with their interests, which plays a pivotal role in mitigating users' information overload problem. Although many recent works have been studied for better personalized news recommendation, the following challenges should be explored more: (C1) Comprehending manifold intents coupled within a news article, (C2) Differentiating varying post-read preferences of news articles, and (C3) Addressing the cold-start user problem. To tackle the aforementioned challenges together, in this paper, we propose a novel personalized news recommendation framework (CROWN) that employs (1) category-guided intent disentanglement for (C1), (2) consistency-based news representation for (C2), and (3) GNN-enhanced hybrid user representation for (C3). Furthermore, we incorporate a category prediction into the training process of CROWN as an auxiliary task, which provides supplementary supervisory signals to enhance intent disentanglement. Extensive experiments on two real-world datasets reveal that (1) CROWN provides consistent performance improvements over ten state-of-the-art news recommendation methods and (2) the proposed strategies significantly improve the accuracy of CROWN.

Problem

• Comprehending manifold intents in news articles
• Differentiating post-read preferences of news articles
• Addressing the cold-start user problem

Innovation

• Category-guided intent disentanglement
• Consistency-based news representation
• GNN-enhanced hybrid user representation