Enhancing Dialogue State Tracking Models through LLM-backed User-Agents Simulation
This work addresses data scarcity for researchers and developers in task-oriented dialogue systems, offering an incremental improvement in data generation efficiency.
The paper tackled the high cost of annotated data for Dialogue State Tracking (DST) by using GPT-4 to simulate user-agent interactions and generate thousands of labeled dialogues, resulting in improved performance over baselines trained only on real data on two public benchmarks.
Dialogue State Tracking (DST) is designed to monitor the evolving dialogue state in the conversations and plays a pivotal role in developing task-oriented dialogue systems. However, obtaining the annotated data for the DST task is usually a costly endeavor. In this paper, we focus on employing LLMs to generate dialogue data to reduce dialogue collection and annotation costs. Specifically, GPT-4 is used to simulate the user and agent interaction, generating thousands of dialogues annotated with DST labels. Then a two-stage fine-tuning on LLaMA 2 is performed on the generated data and the real data for the DST prediction. Experimental results on two public DST benchmarks show that with the generated dialogue data, our model performs better than the baseline trained solely on real data. In addition, our approach is also capable of adapting to the dynamic demands in real-world scenarios, generating dialogues in new domains swiftly. After replacing dialogue segments in any domain with the corresponding generated ones, the model achieves comparable performance to the model trained on real data.