Abstract
In this research, a text processing is assumed to be used for an interface robot. Then, a sentence structure analysis in the text processing is researched and developed here. This paper aims to obtain a solution of a dependency grammar analysis problem when there are two order words or more in certain sentences. The dependency grammar analysis machine focused on functional words (particle, auxiliary verb etc. ) attached to content words was made from useful features of a structural analysis of sentences. Because it was a purpose to establish the model of dependency by a little calculation as a method of deciding modification, a fuzzy pattern matching method was used. The accuracy of a dependency grammar analysis and the correct answer accuracy of sentences are researched by using newspapers as test data for comparison. The number of rules learned as results was 116, the accuracy rate of sentences was about 90%, and the accuracy rate of dependency grammar analysis was about 55%
