Reproduce “Towards Empathetic Conversational Recommender Systems” 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

Xiaoyu Zhang, Ruobing Xie, Yougang Lyu, Xin Xin, Pengjie Ren, Mingfei Liang, Bo Zhang, Zhanhui Kang and 2 more

Researchers propose an Empathetic Conversational Recommender (ECR) framework that integrates emotion-aware components into existing CRSs, leading to more accurate item recommendations and emotionally engaging responses. Experiments show ECR achieved a 6.9% AUC improvement in item recommendation over baselines and generated responses with 73.5% higher emotional intensity and better persuasiveness.

Key insights

  • Incorporating "empathy," defined as capturing and expressing emotions, is crucial for bridging the gap between Conversational Recommender System (CRS) outputs and genuine user needs, fostering more human-like interactions.
  • Modeling user preferences with emotion-aware entity representations (local and global) improves recommendation accuracy by providing a more nuanced understanding of user sentiment.
  • Fine-tuning large language models with rich emotional review data and augmenting responses with retrieved knowledge enables the generation of emotionally engaging, persuasive, and factually grounded responses.

Summary by alphaXiv.

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