A Hybrid Approach to Biomedical Named Entity Recognition and Semantic Role Labeling

Richard Tzong-Han Tsai

Abstract: In this paper, we describe our hybrid approach to two key NLP technologies: biomedical named entity recognition (Bio-NER) and (Bio-SRL). In Bio-NER, our system successfully integrates linguistic features into the CRF framework. In addition, we employ web lexicons and template-based post-processing to further boost its performance. Through these broad linguistic features and the nature of CRF, our system outperforms state-of-the-art machine-learning-based systems, especially in the recognition of protein names (F=78.5%). In Bio-SRL, first, we construct a proposition bank on top of the popular biomedical GENIA treebank following the PropBank annotation scheme. We only annotate the predicate-argument structures (PAS's) of thirty frequently used biomedical verbs (predicates) and their corresponding arguments. Second, we use our proposition bank to train a biomedical SRL system, which uses a maximum entropy (ME) machine-learning model. Thirdly, we automatically generate argument-type templates, which can be used to improve classification of biomedical argument roles. Our experimental results show that a newswire English SRL system that achieves an F-score of 86.29% in the newswire English domain can maintain an F-score of 64.64% when ported to the biomedical domain. By using our annotated biomedical corpus, we can increase that F-score by 22.9%. Adding automatically generated template features further increases overall F-score by 0.47% and adjunct (AM) F-score by 1.57%, respectively.