Supporting Creation of FAQ Dataset for E-Learning Chatbot

Recently, many universities provide e-learning systems for supporting classes. Though the system is an effective and efficient learning environment, it usually lacks a dynamic user support systems. A chatbot is a good choice to support a dynamic QA however, it is difficult to collect the large number of Q&A data or high-quality datasets required to train the chatbot model to obtain high accuracy. In this paper, we propose a novel framework for supporting dataset creation. This framework provides two recommendation algorithms: creating new questions and aggregating semantically similar answers. We evaluated our framework and confirmed that the framework can improve the quality of an FAQ dataset.