LOCAL GRAMMAR BASED AUTO-PREFIXING MODEL FOR AUTOMATIC EXTRACTION IN INDONESIAN CORPUS (Focus on Prefix meN-)

Abstract

Many concerns have been given to morphophonemic phenomena in Indonesian from the perspective of morphology-phonology intersection or even semantics. This paper studies the same phenomenon from Natural Language Processing (NLP) application, and proposes an automatic prefixing model which can later be used to perform NLP tasks such as pattern matching and automatic extraction. The focus of this paper is the phenomena of meN- prefixing in Indonesian. The machine readable linguistic resources built in this paper is expected to transform the linguistic description pertaining to the morphophonemic constraints to an applicable mean for NLP. To accomplish this goal, Local Grammar (LG) approach is employed since this approach is very powerful to describe and formalize a linguistic phenomenon. A corpus processing software, UNITEX, is also employed to apply LG in carrying NLP tasks. It demonstrates how morphophonemic prefixing in Indonesian can be carried out automatically to generate an inflected form machine readable dictionaries as lexical resources and how the resources are used for pattern matching and automatic extraction. The existing linguistic resources from the experiment (machine readable dictionaries, regular expressions and LGGs) are maintainable, open for development and application in a larger corpus, and potential to be used to improve search engines performance for on-line documents in Indonesian.