The Calculation Of Correlation Degree Based On Master-slave's Characteristics Applying In The Build Of Hierarchy System

The semantics and hierarchical identification of Chinese complex sentences is a difficulty for Chinese information processing, what is more, automatic identification for hierarchical relation of Chinese complex sentences is the key step to process it.In this paper, based on master-slave feature correlation calculation, we can process the affiliation of three clauses. By studying arising related or homogeneous words between the clauses, we can calculate the degree of association between sub-clauses, by the way, it is easy to ensure the affiliation of three clauses.Because the main analysis object segmentation result is sub-clause, what is more, non-filling state three clauses are used as corpus in this paper, firstly, this article divides the three clauses. In the process of extracting clauses, pseudo clauses of the filter is the main problem that need to be solved. On the one hand using the punctuation to extract separate sentences, on the other hand, two rule banks are established for filtrating pseudo clauses,thus we can confirm the number of sub-clause finally. And then putting these clauses into the three clauses' table, the final test corpus are randomly selected from the table. When the clauses are divided out, we can extract association characteristics between clauses. By analyzing the clauses, finding that there are three types' connections between clauses: the similar sentence structure, the same word form, the same words concept category. Through the comparison of correlation characteristics, we find out that there is an order between associated features. Through analyzing master-slave sequence, calculating the degree association between sub-clauses is the main focus. Finally, according to the correlation characteristics of master-slave order, calculating the size in correlation between sub-clauses, so as to realize dividing the hierarchical structure of three clauses.The experimental results show that we can divide the hierarchical structure of three clauses by using the method, the accuracy rate reached 85.8%.This also explains the method of three clauses dividing the hierarchical structure is effective and feasible.