Abstract: Term selection is one of the principal procedures in natural language processing. Existing advanced methods allow to construct multiword terms, to form hierarchy of related terms, etc. It provides a high quality of problem solutions where these terms are used. But almost always an expert needs a simple tool to glance a document corpus to reveal the most distinctive features. For this purpose we propose the simple program LexisTerm? for one-word term selection based on a well-known criterion of term specificity. Speaking ‘specificity’ we mean the relation of term frequencies in a given document/corpus and in some gold standard as, for example, a National corpus of document. The program has two options, which give an opportunity. to select both specific terms in an individual document and specific terms for the whole corpus. In the paper we describe this program and demonstrate the results of its work on a real example. The program LexisTerm? is free-share.

Keywords: Natural language processing, term selection, indexing

ACM Classification Keywords: I.2.7 Natural Language Processing

Link:

LEXISTERM – THE PROGRAM FOR TERM SELECTION BY THE CRITERION OF SPECIFICITY

Roque Lopez, Mikhail Alexandrov, Dennis Barreda, Javier Tejada

http://foibg.com/ibs_isc/ibs-24/ibs-24-p01.pdf