DG2: Data Augmentation Through Document Grounded Dialogue Generation

TL;DR

This paper introduces DG2, a data augmentation method for document-grounded dialogue systems using generative models to synthesize diverse dialogues, significantly improving training especially in low-resource scenarios.

Contribution

The paper presents a novel automatic data augmentation technique using a user and agent bot to generate diverse dialogues from documents, enhancing dialog system training.

Findings

Significant performance improvement over traditional augmentation methods.

Effective in low-resource settings for dialog system training.

Generative dialogue model produces diverse, high-quality dialogues.

Abstract

Collecting data for training dialog systems can be extremely expensive due to the involvement of human participants and need for extensive annotation. Especially in document-grounded dialog systems, human experts need to carefully read the unstructured documents to answer the users' questions. As a result, existing document-grounded dialog datasets are relatively small-scale and obstruct the effective training of dialogue systems. In this paper, we propose an automatic data augmentation technique grounded on documents through a generative dialogue model. The dialogue model consists of a user bot and agent bot that can synthesize diverse dialogues given an input document, which are then used to train a downstream model. When supplementing the original dataset, our method achieves significant improvement over traditional data augmentation methods. We also achieve great performance in the low-resource setting.