SENTIMENT ANALYSIS WITH LDA ALGORITHM FOR GOVERNMENT POLICY ANALYSIS USING TWITTER

Abstract

The government needs to formulate public social policy opinion, a source of information to improve performance. People use Twitter to post their views about an object or event. When using this opinion, proper analysis is required to use the information generated for policy decisions. The purpose of this study is to use the Latent Dirichlet Allocation (LDA) algorithm and Twitter social media data obtained in real-time using the API provided by Twitter to classify public comments that determine government policy. The results of the analysis show that people's perceptions of government policy on the opinion on Twitter with latent self-allocations formed into 26 topics with a coherence value of 0.53049and the topic that is often discussed is topic 1 with a percentage score of 8.6%, namely regarding government efforts inequality and access to education, health, employment, and infrastructure also contains information on government policies that facilitate business actors in expanding the MSME market.