Research On Chinese Named Entity Recognition For Information Extraction

Information Extraction’s purpose is to extract information in texts according tospecial needs. Named Entity Recognition is a foundamental and importanttechnology in both information extraction and natural language processing fields.Named Entity is the basic unit in natural language texts. It’s impossible to understandtexts properly without correct recognition of named entity. It plays an important rolein kinds of fields such as Information Extraction, Text Classification, InformationRetrieval, Question Answering Track and so on. In Chinese the process of NamedEntity Recognition is usually divided into two parts: Chinese segmentation andnamed entity tagging. In recent years, this technology has been researched andapplied widely and has shown more and more potential value.In this paper, recognition on personal name, place name and institude name hasbeen researched. The main contents include:1．Chinese Named Entity Recognition has been introduced. The difficulties inpersonal, place and institude name recognition have been exactly analyzed.2．Through Hidden Markov Model and Hiarchical Temperal Memory Model,take experiments on Named Entity Recognition, especially on the recognition ofpersonal name, place name and institude name. Hidden Markov Model has been amature model after years of research and application, and has shown brilliantperformance in efficiency and effects. Hiarchical Temperal Memory Model is anmachine learning model come up recently which imitated the structure and functionsof cortex.3． Based on the results of each experiment, compare and analyze theperformance of Hidden Markov Model and Hiarchical Temperal Memory Model.Traditional Hidden Markov Model doesn’t take full advantage of contecxtinformation in the text, compared with Hiarchical Temperal Memory Model. TheHiarchical Temperal Memory Model this paper introduced has shown proper performance and efficiency. Experimental results show that the Hiarchical TemperalMemory Model can be applied to Chinese Named Entity Recognition field suitably.