互联网已成为一个海量的开放式知识库，其中包含着许多有价值的信息，而网页是互联网信息承载的载体，将 信息结构化成为知识库构建的基础。网页信息不仅包含许多指代词，还含有自身的标题。指代词消解是信息结构化的前提， 综合网页信息具有的一般性和特殊性的特点，本文提出基于多特性融合的代词消解方法研究，能更好地适应网页信息代词的 消解，提高网页信息代词消解的准确率。

The Internet has become a mass of open knowledge base, which contains much valuable information. Web pages are the carriers of information, and the foundation of knowledge base construction. Web information contains many pronouns and titles. The pronouns anaphora resolution is the premise of information structure. The integrated web page information has general and special characteristics. This paper proposes a method of pronouns anaphora resolution based on heterogeneous features fusion, which can better adapt to the web information pronouns anaphora resolution and improve the accuracy.