基于决策树的汉语代词共指消解

提出一种统计与规则相结合的决策树算法进行汉语代词共指消解 ,利用规则过滤掉属性冲突的反例 ,一定程

度上弥补了决策树算法忽略属性关联性的缺点. 采用 Chinese Treebank 作为语料进行测试 ,手工标注其中的共指

关系和特征向量 首先用规则过滤 ,然后采用 C415 决策树算法选择先行语. 实验结果显示 ,消解成功率为

82159 ,其中人称代词和指示代词的成功率分别为 87160 和 75121 .-A total based on decision tree of Chinese pronouns refer to digestion and rules proposed by a statistical decision tree algorithm is a combination of Chinese pronoun refers to a total digestion, using rules to filter out properties of the anti-conflict cases, to a certain extent make up for decision tree algorithm ignores the relevance of attributes shortcomings. using Chinese Treebank as the test corpus, which were marked by hand refers to the relationship and characteristics of the vector　first using the rules filter, and then use the decision tree algorithm chooses to phrase C415. The experimental results show that the success rate of 82,159 percent digestion, in which personal pronouns and the demonstrative pronouns of the success rate was 87.16 thousand percent and 75,121 percent.