An Auto-Encoder Matching Model for Learning Utterance-Level Semantic Dependency in Dialogue Generation

TL;DR

This paper introduces an Auto-Encoder Matching model that effectively captures utterance-level semantic dependencies to generate more coherent and fluent dialogue responses, outperforming baseline models.

Contribution

The novel AEM model uniquely combines auto-encoders and a mapping module to learn semantic dependencies at the utterance level in dialogue generation.

Findings

The AEM model produces responses with higher coherence and fluency.

Experimental results outperform baseline models.

Human evaluations confirm improved response quality.

Abstract

Generating semantically coherent responses is still a major challenge in dialogue generation. Different from conventional text generation tasks, the mapping between inputs and responses in conversations is more complicated, which highly demands the understanding of utterance-level semantic dependency, a relation between the whole meanings of inputs and outputs. To address this problem, we propose an Auto-Encoder Matching (AEM) model to learn such dependency. The model contains two auto-encoders and one mapping module. The auto-encoders learn the semantic representations of inputs and responses, and the mapping module learns to connect the utterance-level representations. Experimental results from automatic and human evaluations demonstrate that our model is capable of generating responses of high coherence and fluency compared to baseline models. The code is available at https://github.com/lancopku/AMM