An automatic lexicon generation for indonesian news sentiment analysis: a case on governor elections in Indonesia

Abstract

Sentiment analysis has been popularly used in analyzing data from the internet.  One of the techniques used is lexicon based sentiment analysis.  Generating lexicon is not an easy process, and lexicon in Bahasa Indonesia is rarely available.  This paper proposes an automatic lexicon generation in Bahasa Indonesia for sentiment analysis purpose.  Experiments were performed using the generated lexicon for doing sentiment analysis on Indonesian political news about the 2018 governor election in three provinces in Indonesia. The conducted experiments show promising results where it can predict the candidate’s rank, the election winner, and the percentage of votes for each candidate with better accuracy than the previous work which used manually generated lexicon.