Conversation Graph: Data Augmentation, Training and Evaluation for Non-Deterministic Dialogue Management

TL;DR

This paper introduces the Conversation Graph (ConvGraph), a graph-based method for augmenting data and evaluating non-deterministic dialogue agents, leading to improved success rates in task-oriented dialogue systems.

Contribution

ConvGraph provides a novel graph-based representation for dialogues that enables data augmentation, multi-reference training, and evaluation for non-deterministic dialogue management.

Findings

Data augmentation with ConvGraph improves success rates by up to 6.4%.

ConvGraph enhances training for non-deterministic dialogue agents.

Evaluation shows ConvGraph's effectiveness across multiple datasets.

Abstract

Task-oriented dialogue systems typically rely on large amounts of high-quality training data or require complex handcrafted rules. However, existing datasets are often limited in size considering the complexity of the dialogues. Additionally, conventional training signal inference is not suitable for non-deterministic agent behaviour, i.e. considering multiple actions as valid in identical dialogue states. We propose the Conversation Graph (ConvGraph), a graph-based representation of dialogues that can be exploited for data augmentation, multi-reference training and evaluation of non-deterministic agents. ConvGraph generates novel dialogue paths to augment data volume and diversity. Intrinsic and extrinsic evaluation across three datasets shows that data augmentation and/or multi-reference training with ConvGraph can improve dialogue success rates by up to 6.4%.