A Chinese Person Name Recognition System Based on Agent-based HMM Position Tagging Model

摘要

An Agent-based HMM Position Tagging (AHPT) model was proposed for Chinese person name recognition. The model unified unknown word identification and person name recognition as a single tagging task. Based on context pattern, special name table and position dependent information, the model could integrate both the internal information and surrounding contextual clues for name entity recognition (NER) under the HMM. The experiment shows that the recall rate and precise rate are respectively 95.11% and 94.02%. The result indicates the application of multi-agent framework can substantially improve the performance of HMM in person name recognition.