IRRGN: An Implicit Relational Reasoning Graph Network for Multi-turn Response Selection

TL;DR

This paper introduces IRRGN, a novel graph network that implicitly models dependencies in multi-turn dialogue response selection, significantly improving reasoning and outperforming previous models, even surpassing human performance on some benchmarks.

Contribution

The paper proposes an Implicit Relational Reasoning Graph Network with URR and ODC components, enhancing reasoning ability and option comparison in multi-turn dialogue response selection.

Findings

Achieves state-of-the-art results on MuTual and MuTual+ datasets.

Surpasses human performance on MuTual dataset.

Significantly improves baseline pretrained language models.

Abstract

The task of response selection in multi-turn dialogue is to find the best option from all candidates. In order to improve the reasoning ability of the model, previous studies pay more attention to using explicit algorithms to model the dependencies between utterances, which are deterministic, limited and inflexible. In addition, few studies consider differences between the options before and after reasoning. In this paper, we propose an Implicit Relational Reasoning Graph Network to address these issues, which consists of the Utterance Relational Reasoner (URR) and the Option Dual Comparator (ODC). URR aims to implicitly extract dependencies between utterances, as well as utterances and options, and make reasoning with relational graph convolutional networks. ODC focuses on perceiving the difference between the options through dual comparison, which can eliminate the interference of the noise options. Experimental results on two multi-turn dialogue reasoning benchmark datasets MuTual and MuTual+ show that our method significantly improves the baseline of four pretrained language models and achieves state-of-the-art performance. The model surpasses human performance for the first time on the MuTual dataset.