A Context-aware Natural Language Generator for Dialogue Systems

TL;DR

This paper introduces a context-aware neural language generator for dialogue systems that adapts to user speech patterns, improving response relevance through sequence-to-sequence models trained on contextual data.

Contribution

It presents a novel, fully trainable neural generator that incorporates preceding dialogue context, enhancing response quality in spoken dialogue systems.

Findings

Significant improvements in automatic metrics

Higher human preference scores

Effective adaptation to user speech patterns

Abstract

We present a novel natural language generation system for spoken dialogue systems capable of entraining (adapting) to users' way of speaking, providing contextually appropriate responses. The generator is based on recurrent neural networks and the sequence-to-sequence approach. It is fully trainable from data which include preceding context along with responses to be generated. We show that the context-aware generator yields significant improvements over the baseline in both automatic metrics and a human pairwise preference test.