Towards role-based filtering of disease outbreak reports

摘要: This paper explores the role of named entities (NEs) in classification disease outbreak report. In annotation schema BioCaster, a text mining system for public health protection, important concepts that reflect information about infectious diseases were conceptually analyzed with formal ontological methodology and classified into types roles. Types are specified as NE classes roles integrated NEs attributes such chemical whether it is being used therapy some disease. We focus on explore different ways to extract, combine use them features classifier. addition, we investigate combination semantic categories disease-related nouns verbs. Experimental results using naive Bayes Support Vector Machine (SVM) algorithms show that: (1) improve performance classification, (2) noun verb contribute substantially improvement classification. Both these statistically significant compared baseline ''raw text'' representation. discuss detail effects each terms accuracy, precision/recall F-score measures task.