A DIK-based question-answering architecture with multi-sources data for Medical Self-Service (KG)

Abstract

Medical data is amplified in terms of speed and capacity in a very fast way, which creates obstacles for users to quickly access valid information. We present a DIK-based Question-Answering Architecture for Medical Self-Service. In addition, we propose a model based on the attention mechanism to extract high-quality medical entity concepts from the Chinese Electronic Medical Records (EMR). Then we modeled the medical data based on the DIK architecture (Data graph, Information graph, and Knowledge graph), construct a Question-Answering model (DIK-QA) for medical self-service that meets the needs of users to quickly and accurately find the medical information they need in massive medical data. Finally, we have realized this approach and applied it to real-world systems. The experimental results on our medical dataset show that the DIK-QA can effectively handle 4W (who/what/why/how) questions, which can help users find the information they need accurately.