Bid Farewell to Seesaw: Towards Accurate Long-tail Session-based Recommendation via Dual Constraints of Hybrid Intents

2025-11-11

AI Summary

In session-based recommendation (SBR), the long-tailed item distribution induces a “trade-off” between recommendation diversity and accuracy—termed the seesaw effect—rooted in session-irrelevant noise prevalent among tail items, which existing methods fail to explicitly identify or constrain. Method: We propose a dual-constraint collaborative optimization framework: (1) attribute-aware spectral clustering maps items into a semantic intent space; (2) a hybrid intent representation is constructed, coupled with an intent-level regularization loss that jointly optimizes both long-tail performance and recommendation accuracy within a unified objective. Contribution/Results: The end-to-end trainable framework significantly improves long-tail metrics (e.g., Tail@20 ↑12.7%) across multiple state-of-the-art models and benchmark datasets, without compromising primary recommendation accuracy. It is the first approach to break the seesaw effect in SBR, establishing new state-of-the-art performance for long-tail recommendation.

Technology Category

• Search and Optimization: Learning to Search
• Data Mining & Knowledge Management: Recommender Systems
• Machine Learning: Semi-Supervised Learning

Application Category

• User Modeling, Personalization and Recommendation: Attacks and countermeasures in recommendation systems
• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• Graph Algorithms and Modeling for the Web: Graph embeddings and representation learning for Web-related graphs

Abstract

Session-based recommendation (SBR) aims to predict anonymous users'next interaction based on their interaction sessions. In the practical recommendation scenario, low-exposure items constitute the majority of interactions, creating a long-tail distribution that severely compromises recommendation diversity. Existing approaches attempt to address this issue by promoting tail items but incur accuracy degradation, exhibiting a"see-saw"effect between long-tail and accuracy performance. We attribute such conflict to session-irrelevant noise within the tail items, which existing long-tail approaches fail to identify and constrain effectively. To resolve this fundamental conflict, we propose extbf{HID} ( extbf{H}ybrid extbf{I}ntent-based extbf{D}ual Constraint Framework), a plug-and-play framework that transforms the conventional"see-saw"into"win-win"through introducing the hybrid intent-based dual constraints for both long-tail and accuracy. Two key innovations are incorporated in this framework: (i) extit{Hybrid Intent Learning}, where we reformulate the intent extraction strategies by employing attribute-aware spectral clustering to reconstruct the item-to-intent mapping. Furthermore, discrimination of session-irrelevant noise is achieved through the assignment of the target and noise intents to each session. (ii) extit{Intent Constraint Loss}, which incorporates two novel constraint paradigms regarding the extit{diversity} and extit{accuracy} to regulate the representation learning process of both items and sessions. These two objectives are unified into a single training loss through rigorous theoretical derivation. Extensive experiments across multiple SBR models and datasets demonstrate that HID can enhance both long-tail performance and recommendation accuracy, establishing new state-of-the-art performance in long-tail recommender systems.

Problem

• Addresses accuracy degradation in long-tail session-based recommendation systems
• Resolves the see-saw effect between long-tail and accuracy performance
• Eliminates session-irrelevant noise within tail items through dual constraints

Innovation

• Hybrid Intent Learning via attribute-aware spectral clustering
• Intent Constraint Loss with diversity and accuracy paradigms
• Dual constraint framework transforming seesaw into win-win performance