Breaking Information Cocoons: A Hyperbolic Graph-LLM Framework for Exploration and Exploitation in Recommender Systems
Qiyao Ma, Menglin Yang, Mingxuan Ju, Tong Zhao, Neil Shah, Rex Ying
TL;DR
Information cocoons in recommender systems hinder exposure to diverse content. HERec introduces a hyperbolic graph‑LLM framework that combines semantic profiles with collaborative signals in hyperbolic space, and builds a Dasgupta‑cost‑driven hierarchical representation to enable user‑adjustable exploration–exploitation. The approach achieves state‑of‑the‑art performance in both utility and diversity, including enhanced tail item recommendations, while providing theoretical gradient insights that preserve hierarchical structure. These contributions offer a principled, scalable path to mitigate information cocoons in large‑scale recommender systems and support dynamic user preferences.
Abstract
Modern recommender systems often create information cocoons, restricting users' exposure to diverse content. A key challenge lies in balancing content exploration and exploitation while allowing users to adjust their recommendation preferences. Intuitively, this balance can be modeled as a tree-structured representation, where depth search facilitates exploitation and breadth search enables exploration. However, existing approaches face two fundamental limitations: Euclidean methods struggle to capture hierarchical structures, while hyperbolic methods, despite their superior hierarchical modeling, lack semantic understanding of user and item profiles and fail to provide a principled mechanism for balancing exploration and exploitation. To address these challenges, we propose HERec, a hyperbolic graph-LLM framework that effectively balances exploration and exploitation in recommender systems. Our framework introduces two key innovations: (1) a semantic-enhanced hierarchical mechanism that aligns rich textual descriptions processed by large language models (LLMs) with collaborative information directly in hyperbolic space, allowing for more nuanced updates that respect the underlying hierarchical structure in user-item profiles; (2) an automatic hierarchical representation by optimizing Dasgupta's cost, which discovers hierarchical structures without requiring predefined hyperparameters, enabling user-adjustable exploration-exploitation trade-offs. Extensive experiments demonstrate that HERec consistently outperforms both Euclidean and hyperbolic baselines, achieving up to 5.49% improvement in utility metrics and 11.39% increase in diversity metrics, effectively mitigating information cocoons. We open-source our model implementation at https://github.com/Martin-qyma/HERec.