MATRAG: Multi-Agent Transparent Retrieval-Augmented Generation for Explainable Recommendations

Researchers led by Sushant Mehta have unveiled MATRAG (Multi-Agent Transparent Retrieval-Augmented Generation), a novel framework that addresses the black-box problem in LLM-based recommendation systems. Unlike traditional recommender systems that provide suggestions without justification, MATRAG employs four specialized agents working in concert: a User Modeling Agent that builds dynamic preference profiles, an Item Analysis Agent that extracts semantic features from knowledge graphs, a Reasoning Agent that synthesizes collaborative and content-based signals, and an Explanation Agent that generates natural language justifications grounded in retrieved knowledge. The framework incorporates a transparency scoring mechanism that quantifies explanation faithfulness and relevance, ensuring users can trust the recommendations.

Extensive testing across three benchmark datasets—Amazon Reviews, MovieLens-1M, and Yelp—demonstrated MATRAG's superiority over leading baselines, achieving a 12.7% improvement in Hit Rate and 15.3% in NDCG (Normalized Discounted Cumulative Gain). Critically, human evaluation revealed that 87.4% of generated explanations were rated as helpful and trustworthy by domain experts. This work establishes new benchmarks for transparent, agentic recommendation systems and provides actionable insights for deploying LLM-based recommenders in production environments where user trust is paramount.