Reproduce “CPED: A Large-Scale Chinese Personalized and Emotional Dialogue Dataset for Conversational AI” 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
Yirong Chen, Weiquan Fan, Xiaofen Xing, Jianxin Pang, Minlie Huang, Wenjing Han, Qianfeng Tie, Xiangmin Xu
The CPED dataset offers a large-scale Chinese multimodal dialogue corpus that integrates stable personality traits, dynamic emotional information, and dialogue acts for each utterance. This resource enables the development of conversational AI capable of more human-like, empathetic, and personalized interactions.
Key insights
- Explicitly incorporating personality, emotion, and dialogue acts as control signals significantly enhances the quality of generated responses, improving emotional correlation and personification capabilities.
- The recognition performance for personality traits varies, with Extraversion and Agreeableness showing higher accuracy than Neuroticism, Openness, and Conscientiousness, suggesting differing manifestations in conversational contexts.
- Emotional states in conversations are highly dynamic, exhibiting significant mutation probabilities between utterances, highlighting the complexity of emotion recognition and the benefit of dialogue-level modeling.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.