Automatic Terminology Extraction Using a Dependency-Graph in NLP

Automatic Terminology Extraction (ATE) is a technique for extracting phrases representing a dataset. This technique is required for translating specialistic books and documents. An existing method focused on the fact that terminologies tend to be composed of two or more single nouns. However, it does not deal with modification relations but only co-occurrence relations among single nouns. Moreover, we have to consider the fact that phrases defined as terminology tend to be explained in another sentence when we propose a novel approach. In this study, we propose a method for extracting terminologies from a dataset considering the modification relations obtained by dependency analysis. In particular, we propose how to extract features enabling us to distinguish whether or not the phrase is terminology from a dependency structure of a sentence.