Reproduce “CARTS: Collaborative Agents for Recommendation Textual Summarization” on OpenResearch

Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.

Universal app for Apple silicon and Intel. Requires macOS 11 or later.

About this paper

Jiao Chen, Kehui Yao, Reza Yousefi Maragheh, Kai Zhao, Jianpeng Xu, Jason Cho, Evren Korpeoglu, Sushant Kumar and 1 more

Walmart Global Tech developed CARTS (Collaborative Agents for Recommendation Textual Summarization), a multi-agent LLM framework designed to generate concise, relevant titles for grouped product recommendations. This framework improves user engagement metrics, achieving uplifts of +0.8% in Click-Through Rate, +6.28% in Add-to-Cart Rate, and +3.78% in Gross Merchandise Value in live A/B tests on an e-commerce platform.

Key insights

  • A multi-agent LLM framework specifically designed for recommendation textual summarization can overcome the limitations of single-pass LLMs in meeting strict length and relevance constraints for item groups.
  • Iterative refinement with specialized feedback agents and an arbitration mechanism significantly boosts the quality and relevance of generated summaries, improving item coverage and adherence to constraints.
  • Applying LLMs in a structured, collaborative agent approach can lead to measurable improvements in real-world e-commerce business metrics, bridging academic advancements with industrial needs.

Summary by alphaXiv.

Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.