Boosted Web Named Entity Recognition via Tri-Training

摘要

Named entity extraction is a fundamental task for many natural language processing applications on theweb. Existing studies rely on annotated training data, which is quite expensive to obtain large datasets,limiting the effectiveness of recognition. In this research, we propose a semisupervised learning approach forweb named entity recognition (NER) model construction via automatic labeling and tri-training. The formerutilizes structured resources containing known named entities for automatic labeling, while the latter makesuse of unlabeled examples to improve the extraction performance. Since this automatically labeled trainingdata may contain noise, a self-testing procedure is used as a follow-up to remove low-confidence annotationand prepare higher-quality training data. Furthermore, we modify tri-training for sequence labeling andderive a proper initialization for large dataset training to improve entity recognition. Finally, we apply thissemisupervised learning framework for person name recognition, business organization name recognition,and location name extraction. In the task of Chinese NER, an F-measure of 0.911, 0.849, and 0.845 canbe achieved, for person, business organization, and location NER, respectively. The same framework isalso applied for English and Japanese business organization name recognition and obtains models withperformance of a 0.832 and 0.803 F-measure.