Dimas Arya, Ilka Zufria

Analisis Sentimen Terhadap Program Kampanye Desak Anies Di X Menggunakan Naïve Bayes

Introduction

Analisis sentimen terhadap program kampanye desak anies di x menggunakan naïve bayes. Analisis sentimen kampanye 'Desak Anies' di X jelang Pilpres 2024 menggunakan Naïve Bayes. Mengungkap opini publik (positif, negatif, netral) dari 1401 komentar dengan akurasi 90%. Memberikan insight strategi.

Abstract

Information technology provides many influences including political, economic, artistic, cultural, social and educational. In 2023 the number of internet users in Indonesia will reach 215.63 million. This increase shows how important social media platforms like X are in public discussions. All attention is focused on political campaigns ahead of the Indonesian presidential election in 2024. This research was conducted to determine public sentiment towards two-way campaigns such as "Desak Anies" in X. Using the Naive Bayes algorithm, researchers divided public sentiment into 3 categories, namely positive, negative, and neutral. This shows public opinion and the effectiveness of campaign strategies. This research analyzes public opinion as much as 1401 comment data. The Naive Bayes algorithm is known to be very good at classifying text, followed by text pre-processing which is useful for cleaning text data so that it can be processed further. Next, TF-IDF is used to extract features. Using the Naive Bayes algorithm, sentiment classification shows the distribution of public opinion towards the "Desak Anies" campaign. The results provide useful suggestions for Amin's team in improving their strategy. The classification results show an accuracy of 90%, precision of 96%, recall of 93%, and F1-score of 95%. With more positive comments than neutral and negative comments.

Review

This paper, "Analisis Sentimen Terhadap Program Kampanye Desak Anies Di X Menggunakan Naïve Bayes," presents a timely and relevant study on public sentiment concerning the "Desak Anies" political campaign on X (Twitter) in the lead-up to the 2024 Indonesian presidential election. Given the substantial number of internet users in Indonesia and the critical role of social media in political discourse, understanding public opinion through sentiment analysis is highly valuable. The authors employ the Naïve Bayes algorithm to classify a dataset of 1401 comments into positive, negative, and neutral categories, utilizing text pre-processing and TF-IDF for feature extraction. The abstract reports impressive performance metrics, with an accuracy of 90%, precision of 96%, recall of 93%, and an F1-score of 95%, indicating a robust classification model. The initial finding of more positive comments than neutral or negative ones offers useful preliminary insights into the campaign's reception. While the reported high performance is commendable, the abstract could benefit from a more detailed justification and explanation of certain methodological choices. For instance, the rationale for selecting Naïve Bayes over other common sentiment analysis algorithms (e.g., SVM, Logistic Regression, or more contemporary deep learning approaches) is not explicitly discussed, which would strengthen the paper's contribution by demonstrating its suitability for this specific problem. Further specifics on the data collection process, such as the timeframe for comment retrieval, the exact keywords used, and any measures taken to ensure the representativeness and reduce bias of the 1401 comments, are also absent. Additionally, a clearer description of the pre-processing steps and the specific variant of the Naïve Bayes algorithm used (e.g., Multinomial, Bernoulli) would enhance transparency and reproducibility. Overall, this research offers a valuable data-driven perspective on public sentiment during a critical political period in Indonesia, demonstrating the potential of machine learning in understanding political dynamics. The study's practical relevance in providing "useful suggestions for Amin's team" is noted, although further elaboration on the nature of these suggestions would be beneficial. To fully realize its scientific contribution, the complete manuscript should expand on the methodological details, including a comparative analysis with other algorithms, a thorough discussion of data collection and its implications, and an exploration of the limitations and generalizability of the findings. Addressing these areas would significantly enhance the paper's rigor and impact within both sentiment analysis and political science domains.

Full Text

Comments

You need to be logged in to post a comment.