Enhancing Performance on Seen and Unseen Dialogue Scenarios using Retrieval-Augmented End-to-End Task-Oriented System

TL;DR

This paper introduces a retrieval-augmented approach for end-to-end task-oriented dialogue systems, enhancing their ability to handle both seen and unseen scenarios by dynamically updating a cache of relevant information.

Contribution

The work proposes a simple cache mechanism combined with a retrieval module to improve flexibility and performance of TOD systems on new and existing dialogue scenarios.

Findings

6.7% improvement in joint goal accuracy over baselines

Effective retrieval from cache enhances dialogue understanding

Compatible with existing pre-trained models

Abstract

End-to-end task-oriented dialogue (TOD) systems have achieved promising performance by leveraging sophisticated natural language understanding and natural language generation capabilities of pre-trained models. This work enables the TOD systems with more flexibility through a simple cache. The cache provides the flexibility to dynamically update the TOD systems and handle both existing and unseen dialogue scenarios. Towards this end, we first fine-tune a retrieval module to effectively retrieve the most relevant information entries from the cache. We then train end-to-end TOD models that can refer to and ground on both dialogue history and retrieved information during TOD generation. The cache is straightforward to construct, and the backbone models of TOD systems are compatible with existing pre-trained generative models. Extensive experiments demonstrate the superior performance of our framework, with a notable improvement in non-empty joint goal accuracy by 6.7% compared to strong baselines.