Recognizing Transliterated Names from Chinese Texts Based on Support Vector Machines and Rules

According to the characteristics of transliterated names in Chinese texts, a method of automatic recognition of Chinese transliterated names combining support vector machines (SVMs) with rules is proposed. The attributes of feature vectors based on characters are extracted. A training set is established and the machine learning models of automatic identification of transliterated names are obtained by testing polynomial Kernel functions; the knowledge cannot be acquired completely if we only use the machine learning model, which will affect the recall. Through careful error analysis, the base of recognition-rules is constructed as post-processing steps to overcome the shortcoming of machine learning model. The results show that the method is efficient for identifying transliterated names from Chinese texts