DiaSynth -- Synthetic Dialogue Generation Framework
Sathya Krishnan Suresh, Wu Mengjun, Tushar Pranav, Eng Siong Chng
2024-10-01
Summary
This paper introduces DiaSynth, a framework designed to create synthetic dialogues that can help train dialogue systems across various topics and applications.
What's the problem?
Many dialogue systems, like chatbots or virtual assistants, struggle because there aren't enough specific datasets of conversations for them to learn from. Existing datasets are either too general or too limited in scope, making it hard for these systems to perform well in real-world situations.
What's the solution?
DiaSynth addresses this issue by generating high-quality, contextually rich dialogues using a Large Language Model (LLM). It creates conversations by simulating different personas and incorporating various conversational styles. This allows DiaSynth to produce tailored dialogues that closely resemble natural human interactions, helping to fill the gap left by traditional data collection methods.
Why it matters?
This research is important because it provides an effective way to generate the dialogue data needed to improve AI systems. By creating realistic synthetic conversations, DiaSynth can enhance the training of dialogue systems, making them more capable and versatile in understanding and responding to human communication.
Abstract
The scarcity of domain specific dialogue datasets across various domains, from academic topics to everyday conversations, limits the development of dialogue systems for various applications. Existing research is often constrained either by dialogue datasets that are too general or by niche domain dialogue datasets whose scale does not match the required scale for training dialogue systems. To address this gap, we introduce DiaSynth - a synthetic dialogue generation framework capable of generating high quality, contextually rich dialogues across a wide range of domains. Our approach differs from existing frameworks by dynamically generating dialogues that incorporate simulated personas, subtopics, and diverse conversational characteristics, using a Large Language Model (LLM) with Chain of Thought (CoT) reasoning to create contextually rich, domain-specific dialogues that closely mimic natural human interactions. DiaSynth produces tailored dialogues that emulate realistic conversations. We perform our experiments by generating synthetic data using different LLMs and few-shot examples from DialogSum and SAMSum. The pretrained language models fine-tuned on the synthetic data outperform the base models by 16.47%, while the comparison between models fine-tuned on in-domain data and synthetic data shows that the synthetic data is able to capture 90.48% of the distribution of the in-domain data. The quality of the data generated also scales with the size of LLMs. These results validate DiaSynth's potential as a robust alternative to traditional data collection methods.