IMQA: A Knowledge Graph Driven Question Answering System for Internal Medicine

The prevention and treatment of internal medicine diseases are a focus of public attention. The emergence of various medical websites makes online consultation a popular trend. Knowledge graph can construct the correlation between different knowledge and provide the data support for question answering systems. Therefore, to assist patients with self-diagnosis, a knowledge graph driven question answering system for internal medicine is constructed in this paper. In the implementation, the ontology schema of internal medicine diseases, which is the basis of knowledge graph construction, is designed. Multiple entities and relationships, extracted from the data on professional healthcare websites, are stored in the Neo4j. After the segmentation of the question, the Aho-Corasick automaton algorithm and semantic similarity are applied to entity recognition. The rule-based approach, an accurate way, is used to identify intentions. This intelligent system, which has high value for the treatment of internal medicine diseases, can effectively solve health issues and reduce the cost of the consultation.