SENTIMENT ANALYSIS ON LGBT ISSUES IN INDONESIA WITH LEXICON-BASED AND SUPPORT VECTOR MACHINE ALGORITHMS

Abstract

Non-heterosexual sexual orientation (LGBT) behavior today is one of the most pervasive issues in Indonesian culture. Because of its domino effect on social stability and physical and mental health, the phenomenon known as lesbian, gay, bisexual, and transgender (LGBT) has always been under scrutiny. The development of LGBT people in Indonesia reflects cultural changes that concern many people. Freedom of speech for LGBT people on social media has many public implications. Observation of this phenomenon gives rise to views of anomalies and discrepancies that have drawn criticism. Various attempts have been made to prevent the movement of LGBT people. However, until now, many still debate the pros and cons of this LGBT movement. The lexicon-based method uses a support vector machine to classify public opinion in TikTok video comments about LGBT issues. The lexicon-based method is used as a weighting method, and the support vector machine method is used as a classification method. The results show that the highest gain in sentiment is neutral, with percentage values of 61%, 56%, 68%, 69%, and 63%. The second is positive sentiment, with percentage values of 27%, 27%, 20%, 20%, and 29%. The rest have negative sentiments. With a relatively high accuracy of the five data sets sequentially at 93%, 89%, 95%, 97%, and 91%. This shows that the majority of Indonesians prefer to ignore the issue.

Keywords

• lexicon-based
• lgbt
• sentiment analysis
• support vector machine