FUNCTIONAL STRUCTURE-ANALYSIS OF RESEARCH ARTICLES SELECTED FROM 3 SPECIALTIES - AUTOMATIC CATEGORY ASSIGNMENT

This paper conducts functional structure analysis of research articles , using the Categories, the author previously proposed as a framework for analysing informational content of research articles. Automatic as signment of the Categories is conducted to the corpus of research arti cles selected from three specialties, using matching of the lexcal-clu e-combination-patterns and stochastic rules. The result is, 86%-91% of Category fixed sentences are assinged correct Categories automaticall y. And through analysis of errors and related studies followings are s uggested to improve efficiency of automatic assignment; (1) to conside r inter-sentences relationship, (2) to adopt syntactic patterns of lex ical clues combination, (3) to introduce the wholistic processing stra tegy, (4) to examine on modification of the Categories, which are need ed content evaluation.