Extracting Terms with EXTra

The identification and extraction of terms play an important role in many areas of knowledge-based applications, such as automatic indexing, knowledge discovery and management, as well as in computational approaches to terminology and lexicography. In this paper, we present EXTra, a tool designed to extract and calculate the degree of termhood of multiword expressions as a function of the statistical distribution of their parts and of the presence of other sub-terms. This work describes EXTra‘s algorithm, and provides the results of its evaluation on a task of term extraction from an Italian corpus of documents belonging to the domain of Public Administration.