Hierarchical Interaction Summarization and Contrastive Prompting for Explainable Recommendations

2025-07-08

AI Summary

Existing interpretable recommendation methods rely on user-item embeddings, which suffer from semantic loss due to dimensionality reduction and sparse interactions; directly feeding such embeddings into large language models (LLMs) for explanation generation is further hindered by their lack of intrinsic semantics, necessitating additional parameterized adapters that exacerbate distortion. To address this, we propose a Hierarchical Interaction Summarization and Contrastive Prompting framework: first, generating structured, semantically rich textual profiles for users and items directly from raw interaction data; then, leveraging contrastive prompting to guide an LLM in producing high-fidelity, verifiable explanations. This is the first work to apply pretrained LLMs directly to semantic summarization and explanation generation from interaction behavior—bypassing the embedding bottleneck entirely. Evaluated on multiple benchmark datasets, our method improves explanation quality (e.g., GPTScore) by 5% over strong baselines and significantly outperforms both automated baselines and real user reviews.

Technology Category

• Natural Language Processing: Summarization
• Machine Learning: Large Multimodal Models (LMMs)
• Data Mining & Knowledge Management: Conversational Systems for Recommendation & Retrieval

Application Category

• User Modeling, Personalization and Recommendation: Large Language Models (LLM) for user modeling and recommendation
• Semantics and Knowledge: Data modeling to support human-machine intelligence, including LLMs agents, intelligent system behavior, explanations, and user-friendly interactions
• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking

Abstract

Explainable recommendations, which use the information of user and item with interaction to generate a explanation for why the user would interact with the item, are crucial for improving user trust and decision transparency to the recommender system. Existing methods primarily rely on encoding features of users and items to embeddings, which often leads to information loss due to dimensionality reduction, sparse interactions, and so on. With the advancements of large language models (LLMs) in language comprehension, some methods use embeddings as LLM inputs for explanation generation. However, since embeddings lack inherent semantics, LLMs must adjust or extend their parameters to interpret them, a process that inevitably incurs information loss. To address this issue, we propose a novel approach combining profile generation via hierarchical interaction summarization (PGHIS), which leverages a pretrained LLM to hierarchically summarize user-item interactions, generating structured textual profiles as explicit representations of user and item characteristics. Additionally, we propose contrastive prompting for explanation generation (CPEG) which employs contrastive learning to guide another reasoning language models in producing high-quality ground truth recommendation explanations. Finally, we use the textual profiles of user and item as input and high-quality explanation as output to fine-tune a LLM for generating explanations. Experimental results on multiple datasets demonstrate that our approach outperforms existing state-of-the-art methods, achieving a great improvement on metrics about explainability (e.g., 5% on GPTScore) and text quality. Furthermore, our generated ground truth explanations achieve a significantly higher win rate compared to user-written reviews and those produced by other methods, demonstrating the effectiveness of CPEG in generating high-quality ground truths.

Problem

• Generating explainable recommendations using user-item interactions
• Overcoming information loss in embedding-based recommendation methods
• Improving explanation quality via contrastive learning and LLMs

Innovation

• Hierarchical summarization for structured user-item profiles
• Contrastive prompting enhances explanation quality
• Fine-tuning LLMs with profiles and high-quality explanations