Data Augmentation by Prompt Tuning on Natural Language Understanding Tasks
Yu-Hao Wang, Chia-Ming Chang, Yi-Hang Tsai, San‐Yih Hwang · 2024
With the advancement of NLP technology, many consumer services now employ chatbots to assist users in obtaining information or providing services. A key component of a chatbot is the natural language understanding module, whose training requires a large amount of data. However, the available training data is often limited. In such cases, data augmentation techniques are employed to generate additional data. In this study, we leverage pre-trained language models and further train them to generate data in the target domain. Specifically, we propose a multi-tasks generation framework. By integrating intent classification and entity recognition tasks, the generated data can be used for multi-task training, and it demonstrates that such integration can enhance the performance of both tasks.