Biomedical named entity recognition based on multistage three-way decisions

Abstract

Biomedical named entity recognition (Bio-NER) is one of the most fundamental tasks in the field of biomedical information extraction. The accuracy of biomedical named entity recognition is crucial to the follow-up research work. This paper presents a method for named entity recognition based on the concept of three-way decisions. The method uses a discriminative approach named conditional random fields (CRFs) to construct models. These models follow the decision-making rule of three-way decision in all stages, the model cannot make decision arbitrarily when the information is incomplete until it gets more information. The experimental results show that our method can improve the performance for biomedical named entity recognition compared with other methods.