HeterMPC: A Heterogeneous Graph Neural Network for Response Generation in Multi-Party Conversations

TL;DR

HeterMPC introduces a heterogeneous graph neural network that models multi-party conversation structures to improve response generation, outperforming existing models on IRC benchmarks.

Contribution

This paper presents HeterMPC, a novel heterogeneous graph neural network that captures complex interactions in multi-party conversations for response generation.

Findings

HeterMPC outperforms baseline models on IRC dataset.

The model effectively captures speaker and utterance semantics.

Multi-hop updating enhances response relevance.

Abstract

Recently, various response generation models for two-party conversations have achieved impressive improvements, but less effort has been paid to multi-party conversations (MPCs) which are more practical and complicated. Compared with a two-party conversation where a dialogue context is a sequence of utterances, building a response generation model for MPCs is more challenging, since there exist complicated context structures and the generated responses heavily rely on both interlocutors (i.e., speaker and addressee) and history utterances. To address these challenges, we present HeterMPC, a heterogeneous graph-based neural network for response generation in MPCs which models the semantics of utterances and interlocutors simultaneously with two types of nodes in a graph. Besides, we also design six types of meta relations with node-edge-type-dependent parameters to characterize the heterogeneous interactions within the graph. Through multi-hop updating, HeterMPC can adequately utilize the structural knowledge of conversations for response generation. Experimental results on the Ubuntu Internet Relay Chat (IRC) channel benchmark show that HeterMPC outperforms various baseline models for response generation in MPCs.