Analisa Sentiment Masyarakat terhadap Pemilu 2019 berdasarkan Opini di Twitter menggunakan Metode Naive Bayes Classifier

Abstract

Abstrack-Many people expressed their opinions on the 2019 Election via Twitter. The habit of people posting tweets to assess candidates for election is one of the media that represents the community's response to candidates. Approaching the general election, usually there are certain parties who want to know the sentiments and responses to the candidate figures such as the survey institution. However, a system is very difficult to measure people's tendencies towards candidates because an opinion has a free language or a diverse expression language. We propose a new approach in the analysis of opinions about elections. This is opinion mining based on time. This study proposes an approach that will extract and process textstual data automatically to obtain sentiment information contained in opinion sentences. There are several stages for conducting sentiment analysis, including the data collection stage, the data preprocessing stage, the opinion sentiment analysis stage by using the naive bayes classifer method, then the last is the result visualization stage. This approach is time-based, so opinion analysis can be displayed based on time intervals. The results of this study indicate that the results of the accuracy of the naïve bayes classifier method are 62% with a recall value of 45% and precision is 41%.