Corpus Construction for Named-Entity and Entity Relations for Electronic Medical Records of Cardiovascular Disease

AbstractElectronic medical record (EMR) is an important carrier of medical health information and contains rich knowledge in medical field. Based on the text of EMR of CardioVascular Disease (CVD) and referring to the existing relevant work at home and abroad, this paper formulated the classification system of entities and entity relations and labeling norms suitable for this labeling. The corpus of CardioVascular Disease Electronic Medical Record (CVDEMRC) entity and entity relation is constructed semi-automatically by constructing sentence dictionary and multiple rounds of manual review annotation. The constructed CVDEMRC contains 7,691 named entity concepts and 11,185 entity relations, and both entity and relation have achieved high consistency results, which can provide data basis for CVD research.KeywordsCardiovascular diseaseCorpus constructionNamed entityEntity relations