Syntactic analysis is located in a core position in natural language processing, its performance has important influence to other technology. In grammar, syntax interdependence with the form concise, easy to mark or facilitate application advantages, gradually to the attention of the researchers. With the development of natural language and based on statistical machine learning, dependent syntax analysis algorithm based on statistics is more and more close to maturity. And the lack of interdependence syntactic tree of Chinese library at present stage limit the development of Chinese interdependence syntactic analysis. The size of the corpus of the accuracy determines the syntactic analysis of the performance, for dependent syntax analysis based on tree library, without the large-scale, high accuracy corpus, the algorithm will lost the role. Tree library as a kind of sentence for the deep syntactic tagging corpus increasingly aroused people’s concern.This paper describes the algorithm and related statistical machine learning model dependency parsing and analysis summarizes the advantages and disadvantages of these algorithms; Then according to Chinese dependency syntactic features to make marked norms, so that the resulting library dependency tree is more compliant to Chinese language, and has practical value; Secondly, establish a more sophisticated rules for the special structure of Chinese dependency syntax; Once more, built a phrase structure tree into a dependency tree library system; Finally, experiments confirmed the newborn idioms material advantages, as well as research corpus own characteristics, found factors of influence Chinese dependency parsing, and from the perspective of the tree library itself, research to method of improving the accuracy of dependency parsing. |