Title:DRE: Generating Recommendation Explanations by Aligning Large Language Models at Data-level

Abstract:Recommendation systems play a crucial role in various domains, suggesting items based on user this http URL, the lack of transparency in presenting recommendations can lead to user confusion. In this paper, we introduce Data-level Recommendation Explanation (DRE), a non-intrusive explanation framework for black-box recommendation this http URL from existing methods, DRE does not require any intermediary representations of the recommendation model or latent alignment training, mitigating potential performance this http URL propose a data-level alignment method, leveraging large language models to reason relationships between user data and recommended this http URL, we address the challenge of enriching the details of the explanation by introducing target-aware user preference distillation, utilizing item reviews. Experimental results on benchmark datasets demonstrate the effectiveness of the DRE in providing accurate and user-centric explanations, enhancing user engagement with recommended item.