Chinese Named Entity Recognition Based On Conditional Random Fields

Named entity recognition is one of the fundamental problems in many natural language processing applications, such as information extraction, information retrieval, machine translation, shallow parsing and question answering system. The research of named entity recognition is of great worth.According to the modern Chinese characteristics, this paper mainly researches Chinese named entity recognition including person names, location names and organization names. We design and implement a Chinese named entity recognition system based on conditional random fields.This paper is organized as follows:First, it introduces the difficulties of named entity recognition and the characteristics of person names, location names and organization names. It also compares various named entity recognition methods and some existing Chinese named entity recognition systems.Then this paper introduces the definition of conditional random fields, the graph structure, potential functions, parameters estimation and probability computations. Regarding conditional random fields as the basic frames, this paper proposes different feature templates for different kinds of named entities.Finally, it presents a cascaded Chinese named entity recognition system based on conditional random fields. In the system, person names, simple location names and simple organization names are recognized by the lower model at first, and then the result of the lower model is passed to the high model for recognizing the complex location names and organization names. The experimental results show that the system has achieved good performance. In the open test, the recall, precision and F-measure has reached 82.50%, 76.04% and 79.14%, respectively.