Chinese Named Entity Recognition with Conditional Random Fields in the Light of Chinese Characteristics

This paper introduces the research works of Chinese named entity recognition (CNER) including person name, organization name and location name. To differ from the conventional approaches that usually introduce more about the used algorithms with less discussion about the CNER problem itself, this paper firstly conducts a study of the Chinese characteristics and makes a discussion of the different feature sets; then a promising comparison result is shown with the optimized features and concise model. Furthermore, different performances are analyzed of various features and algorithms employed by other researchers. To facilitate the further researches, this paper provides some formal definitions about the issues in the CNER with potential solutions. Following the SIGHAN bakeoffs, the experiments are performed in the closed track but the problems of the open track tasks are also discussed.