English

GSN: A Graph-Structured Network for Multi-Party Dialogues

Computation and Language 2019-06-03 v1 Machine Learning

Abstract

Existing neural models for dialogue response generation assume that utterances are sequentially organized. However, many real-world dialogues involve multiple interlocutors (i.e., multi-party dialogues), where the assumption does not hold as utterances from different interlocutors can occur "in parallel." This paper generalizes existing sequence-based models to a Graph-Structured neural Network (GSN) for dialogue modeling. The core of GSN is a graph-based encoder that can model the information flow along the graph-structured dialogues (two-party sequential dialogues are a special case). Experimental results show that GSN significantly outperforms existing sequence-based models.

Keywords

Cite

@article{arxiv.1905.13637,
  title  = {GSN: A Graph-Structured Network for Multi-Party Dialogues},
  author = {Wenpeng Hu and Zhangming Chan and Bing Liu and Dongyan Zhao and Jinwen Ma and Rui Yan},
  journal= {arXiv preprint arXiv:1905.13637},
  year   = {2019}
}
R2 v1 2026-06-23T09:35:24.575Z