OPTIMASI ALGORITMA STEMMING PORTER UNTUK PEMROSESAN TEKS DALAM BAHASA INDONESIA

Abstract

Stemming is a method for finding root words by removing affixes. This process is crucial in data processing because it affects system performance. Numerous stemming algorithms exist, including the Porter algorithm. However, these algorithms often fail in the stemming process, necessitating rule development. The proposed development involves three branching paths, each of which checks the word in a dictionary. The goal is to improve performance and reduce errors. This study tested stemming failures from previous studies with an optimized algorithm that improves the process from the Porter library. This optimization eliminates the possessive pronoun stage and always checks for stemmed words. Furthermore, testing was conducted using data from Twitter with the keyword KOMINFO. The results showed an accuracy of 84% for the stemming optimization path, better than the 80% accuracy of the Porter algorithm.

Keywords: Communication and Information Technology, NLP, Porter, Stemming, Text Preprocessing