A Pre-Identification Method for Chinese Named Entity Recognition

摘要

In this paper, a pre-identification method forChinese named entity recognition is proposed. Internalinformation of entity name like family name, first name inperson name, feature word in place name and organizationname do not needed. Through entity name guessing basedon context keywords, pre-identification is realized.Definition of bidirectional potential entity name recognition,rough confirmation of potential entity name, segmentationword is proposed. To solve the possible ambiguity in entityname identification, the degree of segmentation andconjunction is presented as well as cascade recognition andfinal confirmation. Combining with this pre-processingmethod, performance will be improved by using internalinformation of entity name. Experiment proves that themethod have a special advantage in recognition specialentity name, ambiguity name and irregular name. In thispaper, Chinese person name is taken as an example forentity name recognition. Nevertheless, the method is notlimit to person name recognition but also a preidentificationmethod for other entity name.