A New Approach of Question Answering based on Knowledge Graph in Traditional Medicine

Abstract

In recent years, it has been great interest for Question Answering (QA) systems applied to many areas placing a high value on the community.The study and development of such QA systems through chatbot tools in medicine raise great needs for clinicians in their daily activities.Chatbots use the knowledge that could be retrieved from a database, but with limited inference capability.In this paper, we propose a new QA system based on Knowledge Graph (knowledge graph) for Traditional Medicine.Data of the knowledge graph is obtained from two sources including those from diagnostic of treatment diagrams and those collected on well-known medical websites through the Internet.The knowledge graph is then formed by combining the entities and relationships using the Named Entity Recognition (NER) model.Diagnosis is made via the node similarity algorithm in the knowledge graph for symptom identification.The effectiveness of the system is demonstrated through theoretical analysis and real-world experimental outcomes.