Chinese Person Named Entity Recognition Using Wikipedia and Hidden Markov Model

• Main Authors: Chen PO-Hung, 陳柏宏
• Other Authors: Huang Chuen-min
• Format: Others
• Language: zh-TW
• Published: 2015
• Online Access: http://ndltd.ncl.edu.tw/handle/25020303867299358162

碩士 === 國立雲林科技大學 === 資訊管理系 === 103 === There are many researches that using machine learning to recognize Chinese name entity, but those researches used news or existing corpus. E.g. IEER-99、MET2. With different culture evolution and time passing, name rule has been changing and increasing. Maintain the name database become big problem. Therefore, this study proposed that the use of Wikipedia for training data to recognize the Chinese name. Combining the data source which extract from the Wikipedia and Hidden Markov Model (HMM) to solve the problem. In order to validate the proposed method are valid, we carried out a two-stage evaluation method. Experimental results show that the proposed method can effectively identify different name which generate by different rule. The advantage to use Wikipedia is solving the problem which need to collect large training data and the maintenance of Wikipedia is in real time. Phase I use cross-validation and the precision, recall and rates F-Measure of the Phase I are 87%, 79% and 83%, respectively. Phase II is to identify the students name in 2012 Taiwan General Scholastic Ability Test. The precision, recall and rates F-Measure of evaluation are 95%, 91% and 93%