基于关联规则的中文姓名识别方法

Chinese Name Recognition based on Association Rules

从中文姓名识别技术是中文分词领域的重要分支的角度,分析了传统统计中文切词方法正确率低的缺失,表现为基于机器学习方法的中文姓名识别技术对词典规模和语料库容量依赖性大的缺陷,以至于不能够有效地发现未登录词,降低了中文姓名识别的正确率;以中文姓氏用字为线索,提出了一种改善的无监督的中文姓名识别方法,在统计法的基础上,利用关联规则算法挖掘出与姓氏相关的强规则进行姓名识别。实验结果表明:此方法对词典规模和语料库依赖性较小,能够自动消除歧义,有效地发现未登录词,提高了中文姓名识别的正确率。

Chinese name recognition is an important branch in the field of Chinese phrases .The general methods of it are based on the statistics and the study through machine.The method of statistics has a low currency of problems,and there is great dependence by the way of the study of machine.According to these reasons,the author presents other Chinese name recognition methods which can solve these problems,that is,finding out the impulsive rules to recognize Chinese names through relative rules algorithm.The experiments indicate that this method can reduce the dependence on the scope of dicitionaries and the corpus,eliminate the ambiguity and find out the words effectively that are not registered,so the currency of Chinese name recognition can be improved.