Text mining for information screen in risk assessment of environmental endocrine disruptive chemicals

Abstract

Text mining technique has been widely applied in numerous fields of research, among which, the employment of machine learning in biological text analysis and management has increased its popularity in recent years. In this study, machine learning based automatic classification system has been constructed according to the requirement of the environmental health risk assessment for endocrine disruptive chemicals, as the knowledge explosion has made traditional manual assessment impossible. And the factors that may influence the classifier training and performance were compared and selected. The constructed classifier has been proved to be with high accuracy and efficiency, which has important significance on computer based risk assessment for various potential hazardous chemicals. © 2014 Springer-Verlag Berlin Heidelberg.