Study Comparison Stemmer to Optimize Text Preprocessing In Sentiment Analysis Indonesian E-Commerce Reviews

Stemming was an essential part of text preprocessing. Stemming serves to map various forms of words into crucial words. Many stemming algorithms have been developed, including Indonesian stemmers. In this study, we will compare the rule-based stemmer algorithm and dictionary on sentiment analysis classification in the Indonesian E-Commerce Review. The purpose of this research is to find out the best preprocessing algorithm for K-Nearest Neighbor (KNN) classification. The results of the two stemmer testing still contain errors. KNN shows that the use of rule-based stemmers has an effect on increasing accuracy than without using a stemmer. In comparison, the dictionary-based method does not have much influence on the performance of the KNN method.