Analysis of Indonesia Politics Polarization before 2019 President Election Using Sentiment Analysis and Social Network Analysis

Abstract

The development of the Internet in Indonesia is quite rapid, it is marked by the increasing use of social networks, especially Twitter. Not only to share status or stories, Twitter has become become a means of promotion and campaign for elections. The Twitter data can be used to find out the political polarization in Indonesia that is needed in the 2019 presidential election. The method used in this research is sentiment analysis using naï ve bayes classifier and social network analysis using the calculation of network attribute values and centrality values. 8.814 Twitter data was collected using data crawling method. The data are divided into three subsets consisting of jokowi's sentiments, prabowo's sentiments, and pilpres's sentiments. Final result of the sentiment analysis is classified sentiments into positives, negatives, and neutral. The average value of the classification results was 91.27% positive sentiment, 7.56% negatives sentiment, and 1.17% neutral sentiment. This classification yielded the average accuracy of 69.2% for jokowi's sentiments and 100% for prabowo sentiments. The classification accuracy calculation uses ROCs method. Final results of the social network analysis based on the calculation of network attributes yielded 277 nodes, 7.950 edges, 57,401 average degree, 56.44 average weighted degree, network diameter is 4, 1.853 average path length, 0.201 density, and 5 of number communities. Centrality values generates the 5 most influential actors in social network interactions are jokowi's of first rank, 2 nd SBYudhoyono's, 3 rd detikcom, 4 th yjuniardi, 5 th mohmahfudmd.

ERIS: An Approach Based on Community Boundaries to Assess Polarization in Online Social Networks

“…Many works on social polarization use exploratory analyzes combining metrics from the graph theory with interpretations provided by Natural Language Processing (NLP) tools like sentiment analysis based on Naïves Bayes [2,20] or sentence embedding models based on Retweet-BERT [22]. These approaches best capture the semantics of discussions but require heavy involvement from the analyst, especially during the preprocessing step.…”

Analisis Penyebaran Informasi Vaksin Covid-19 Pada Twitter Menggunakan Kolaborasi SNA dan Sentiment Analysis

“…Klasifikasi SVM bertugas mencari hyperplane terbaik guna memisahkan kelas bersentimen positif (+1) dan kelas bersentimen negatif (-1) [11]. Penelitian terkait SNA dan Sentiment Analysis sudah banyak dilakukan, penelitian dengan judul Analysis of Indonesia Politics Polarization before 2019 President Election Using Sentiment Analysis and Social Network Analysis menunjukan bahwa penelitian yang dilakukan untuk mengklasifikasikan sentimen serta mengevaluasi dengan memodelkan penyebaran informasi yang terjadi di jejaring sosial Twitter [12]. Penelitian lain dengan judul Ekstraksi Knowledge Tentang Penyebaran #RATNAMILIKSIAPA Pada Jejaring Sosial (Twitter) Menggunakan Social Network Analysis (SNA) dilakukan untuk menemukan aktor-aktor yang yang berpengaruh dalam jaringan terhadap kelompok-kelompok yang terbentuk.…”

Analisa Interaksi Pengguna Media Sosial Perusahaan Sekuritas di Indonesia Saat Covid-19 menggunakan Social Network Analysis (Studi Kasus: Indopremier dan Bursa Efek Indonesia)

“…Oleh karena itu, individu harus gunakan unit sosial dasar. Masyarakat (atom sosial) terdiri dari individu dan hubungan sosial, ekonomi, atau budaya dan diskusi menjadi kelompok dan akhirnya terdiri dari kelompok yang saling terkait dijelaskan dalam sosiogram (struktur hubungan antar kelompok) [6]. Terdapat lima properti jaringan dalam jejaring sosial, antara lain : Tabel 1.…”