一种汉语语句依存关系网分析策略与生成算法研究

本文以依存语法作为语言模型的基础,首先提出了一种全新的句子分析策略: 分析在两级上进行. 一级是基于中心动词同其他成分间的约束关系 ,通过寻找汉语语义类之间可能存在的句法关系,实现句子成分过滤,完成句子主干提取. 另一级是上下文级,将语法、语义和语境信息一体化,结合依存语法确定汉语句子中各成分间的依存关系. 之后给出了一种快速有效的用于完成句子分析的松弛匹配迭代算法. 通过实验表明了该分析策略和算法的可行性.

A study of the parsing strategy and the generative algorithm for dependency relation network of Chinese sentences.

Taking dependency grammar as language processing model, this paper proposed a new strategy of Chinese sentence component analysis. Strategy of parsing works on sentence level and context level.On sentence level,a sentence consititute filtering algorithem was implemented to extract sentence frame. On context level,the syntactic and semantic information were complatible used to obtain the dependency relations between different components based on dependency grammar. Then a faster relaxation algorithm of parsing was given. The result of experiment shows that this strategy and algorithm is feasible not only in theory but also in practice.