Response Selection for Multi-Party Conversations with Dynamic Topic Tracking

TL;DR

This paper introduces a novel framework for response selection in multi-party conversations by framing it as a dynamic topic tracking task, utilizing a new pretraining method called Topic-BERT, achieving state-of-the-art results.

Contribution

It proposes a multi-task learning framework with Topic-BERT for dynamic topic disentanglement and response selection in multi-party conversations, addressing limitations of existing two-party focused methods.

Findings

Achieves state-of-the-art response selection accuracy on DSTC-8 dataset.

Effectively disentangles multiple conversation topics.

Outperforms existing methods by a significant margin.

Abstract

While participants in a multi-party multi-turn conversation simultaneously engage in multiple conversation topics, existing response selection methods are developed mainly focusing on a two-party single-conversation scenario. Hence, the prolongation and transition of conversation topics are ignored by current methods. In this work, we frame response selection as a dynamic topic tracking task to match the topic between the response and relevant conversation context. With this new formulation, we propose a novel multi-task learning framework that supports efficient encoding through large pretrained models with only two utterances at once to perform dynamic topic disentanglement and response selection. We also propose Topic-BERT an essential pretraining step to embed topic information into BERT with self-supervised learning. Experimental results on the DSTC-8 Ubuntu IRC dataset show state-of-the-art results in response selection and topic disentanglement tasks outperforming existing methods by a good margin.