Analysis of Twitter Sentiment on the Implementation of Regional Elections in Indonesia During Covid-19 Using the Support Vector Machine Method

Abstract

Sentiment analysis or opinion mining is a series of problem solving based on public opinion. The opinion is in the form of text or writing in the form of documents obtained from social media. Sentiment analysis serves to determine public opinion in responding to a policy, activity or issue that is happening and being discussed, one of which is on Twitter social media. Sentiment analysis in this study focuses on the activities of the 2020 regional elections during the Covid-19 pandemic which was held on 9 December 2020. Twitter social media works in real-time, so in retrieving research data using the Trending Topic feature to retrieve research datasets. The results of the dataset are then processed using text mining techniques and used as material for analysis to determine the public's response to the implementation of the elections during covid- 19 whether it tends to have a positive or negative sentiment, as well as knowing the opinion factors that often arise. The adoption of the Support Vector Machine (SVM) method for sentiment analysis was carried out by testing the composition of various datasets. From the test results using 4 scenarios of training data and test data, namely 90:10, 80:20, 70:30, 60:40, it is obtained that the SVM method can be implemented with an accuracy value of 87% in the data scenario of 80% training data and 20% test data. Variables that affect accuracy are the amount of data, the ratio of the number of training and test data and the ratio of the number of positive and negative data used.