Sentiment Analysis Based on Text About President Candidate 2024 in Indonesia Using Artificial Intelligence with Parameter Optimization Algorithm

• Published in: 2023 IEEE International Conference on Internet of Things and Intelligence Systems (IoTaIS) pp. 216 - 222
• Main Authors: Wulandari, Dian Rezky, Murti, Muhammad Ary, T.S.P, Hilman Fauzi
• Format: Conference Proceeding
• Language: English
• Published: IEEE 28.11.2023
• Subjects: artificial intelligence; Blogs; election prediction; Sentiment analysis; Social networking (online); Support vector machines; survey method; Surveys; text processing; Training data; Voting
• ISSN: 2832-1383
• DOI: 10.1109/IoTaIS60147.2023.10346038

Indonesia as a democratic country, is set to hold the election for choose the next President and Vice President every five years. As is customary during election periods, political campaigns are conducted using various methods, ranging from traditional to modern, including social media. This has resulted in a significant increase in the use of social media platforms. As a result, social media data has become a valuable source of information, including public opinion about the candidates. The purpose of the research is to make the improvement for sentiment analysis on election prediction in Indonesia using Artificial Intelligence and optimization algorithm. Also, the purposed of this research is to make the approach of do the survey about the public response towards president candidate so it can improve their engagement and outreach their strategies. The sentiment analysis does in the research is take the data from Twitter using several keywords and it will be labeled using Valence Aware Dictionary for Sentiment Reasoning (VADER) method and the performance of it will be test using three algorithms such as Naive Bayes, K - Nearest Neighbor (KNN) and Support Vector Machine (SVM). The result of research show that VADER method combined with SVM and Grid Search CV for parameter optimization get the best result where it gets 91.93% of accuracy, 91.76% of recall, 97.94% of precision and 91.84% of f1 - score with test size 0.1, c value 0.1 and RBF kernel for SVM Parameter. Compared with traditional surveys, this proposed method can provide the first understanding of preference patterns based on data and also can be an additional tool for analyzing the preferences of presidential candidates.