Construction of Disease-Symptom Knowledge Graph from Web-Board Documents

The research aim is to construct a disease-symptom knowledge graph (DSKG) as cause-effect containing relations relation type determined from downloaded documents on medical web-board resources. Each connects disease-name concept node (a causative-concept node) corresponding having group of correlated symptom-concept/effect-concept features common among some concepts. DSKG benefits non-professionals in preliminary diagnosis through recommender web-board. There are three main problems: how determine symptom concepts sentences without annotation the documents’ topic-names; with/without complications; and involving high dimensional symptom-concept after union groups. Therefore, we apply word co-occurrence pattern including medical-symptom expressions Wikipedia MeSH Lexitron Dictionary Cartesian product applied for automatic-supervised machine learning relation. We propose using Principal Component Analysis constructing by dimensionality reduction with minimized information loss. In contrast previous works, proposed approach enables construction precise concise representation scores 7.8 9, respectively.