收录:
摘要:
Named entity recognition is a basic and core task of biomedical text mining. Comparing with other named entity recognition methods, methods based on domain relevance measurement need the smaller training corpora and entity samples and are appropriate for recognizing narrow-domain entities, which belong to a subdivision and small semantic class. However, how to obtain the high-quality target corpus set become a key issue. This paper propose a biomedicine named entity recognition method by integrating domain contextual relevance measurement and active learning. Firstly, binding with densitybased clustering and semantic distance measurement, the representative and informative contexts are selected to construct the target corpus set by an active learning approach. Secondly, the domain contextual relevance of candidate entities is calculated by using Domain the discrimination degree and domain dependence function for recognizing the target entities. Experimental results show that the proposed method can effectively reduce training time and improve the accuracy of entity recognition. © 2019 IEEE.
关键词:
通讯作者信息:
电子邮件地址: