Adversarial Learning on the Latent Space for Diverse Dialog Generation

TL;DR

This paper introduces a two-step adversarial framework for dialog response generation that improves relevance, fluency, and diversity by learning meaningful sentence representations and mapping queries to responses.

Contribution

The paper presents a novel two-step generative adversarial network approach that enhances response diversity and relevance in dialog systems.

Findings

Generated responses are more fluent and relevant.

Model produces more diverse responses than existing methods.

Quantitative and qualitative evaluations confirm improved performance.

Abstract

Generating relevant responses in a dialog is challenging, and requires not only proper modeling of context in the conversation but also being able to generate fluent sentences during inference. In this paper, we propose a two-step framework based on generative adversarial nets for generating conditioned responses. Our model first learns a meaningful representation of sentences by autoencoding and then learns to map an input query to the response representation, which is in turn decoded as a response sentence. Both quantitative and qualitative evaluations show that our model generates more fluent, relevant, and diverse responses than existing state-of-the-art methods.