The identification of Chinese named entity in the field of medicine based on Bootstrapping method

In the process of medicine information extraction, there are many Named Entity (NE) need to be recognized But currently the research on identification of NE in the field of medicine, such as physician, hospital, disease, medicine NE etc. is rarely. So in this paper we present an new approach for Named Entity Recognition (NER) in the field of medicine based on Bootstrapping method This method primarily consists of two key steps. The first step is Bootstrapping training procedure. Introduction of Bootstrapping, a self-expanding technology, make this method get rid of human supervision. The idea is to start from a initial feature set extracted from contextual information of a little annotated corpus that corresponding to the concept for the target NE type, and successively form a complete feature set of NE. The other is NER procedure. The feature set of NE adopted from Bootstrapping training procedure is expressed as class feature vector (CFV). Contextual information of candidate NE is expressed as example feature vector (EFV). The degree of similarity of CFV and EFV determines whether the candidate is a NE. We carried out several experiments using corpus from the internet for physician and disease NER The F-measure of physician and disease NER on the test data is 0.926 and 0.969 respectively. The result above show that the Bootstrapping method has the key advantage of NER.