Semi-Supervised Knowledge-Grounded Pre-training for Task-Oriented Dialog Systems

TL;DR

This paper introduces a semi-supervised, knowledge-grounded pre-training approach for task-oriented dialogue systems, leveraging both labeled and unlabeled data to improve performance on a large-scale Chinese dataset, achieving top results in a challenge.

Contribution

The paper presents a novel semi-supervised pre-training method for knowledge-grounded TOD systems, utilizing large-scale unlabeled data to enhance response generation.

Findings

Achieved first place in the SereTOD 2022 challenge.

Significant improvements in BLEU (+7.64) and Success (+13.6%) metrics.

Effective use of semi-supervised learning on real-world Chinese data.

Abstract

Recent advances in neural approaches greatly improve task-oriented dialogue (TOD) systems which assist users to accomplish their goals. However, such systems rely on costly manually labeled dialogs which are not available in practical scenarios. In this paper, we present our models for Track 2 of the SereTOD 2022 challenge, which is the first challenge of building semi-supervised and reinforced TOD systems on a large-scale real-world Chinese TOD dataset MobileCS. We build a knowledge-grounded dialog model to formulate dialog history and local KB as input and predict the system response. And we perform semi-supervised pre-training both on the labeled and unlabeled data. Our system achieves the first place both in the automatic evaluation and human interaction, especially with higher BLEU (+7.64) and Success (+13.6\%) than the second place.