Biomedical name recognition: A machine learning approach

Abstract

The purpose of this research is to automatically recognize names in biomedical documents. First, we analyze characteristics of biomedical names. Then, we propose a rich set of features, including orthographic, morphological, part-of-speech and semantic trigger features. All these features are integrated via a Hidden Markov Model with back-off modeling. Finally, we propose a method for biomedical abbreviation recognition and two methods for cascaded name recognition. Evaluation on GENIA corpus V3.02 and V1.1 shows our system achieves 66.5 and 62.5 F-measure respectively. It shows that our system outperforms previous best published system on the same V1.1 by 8.1 F-measure. The major contribution of this thesis lies in its detailed analysis of biomedical names, the rich feature set and the effective methods for cascaded name recognition. To our best knowledge, our system is the first one that resolves the phenomena of cascaded biomedical names. In addition, a demo has been put on the web.