Multitek Indonesia : Jurnal Ilmiah
Vol 16, No 2 (2022): Desember

ANALYSIS SENTIMENTS IN FACEBOOK DOWN CASE USING VADER AND NAIVE BAYES CLASSIFICATION METHOD

Ilham Firman Ashari (Institut Teknologi Sumatera)



Article Info

Publish Date
31 Dec 2022

Abstract

Abstrak Facebook adalah media sosial terbesar di dunia. Semua aplikasi media sosial besutan Facebook tidak bisa diakses secara bersamaan dalam waktu kurang lebih 6 jam. Hal ini tidak hanya terjadi di Indonesia, tetapi di seluruh negara di dunia, pada tanggal 5-6 Oktober waktu Indonesia. Dengan adanya kasus ini, berbagai komentar dan opini dari masyarakat di Twitter terkait kasus Facebook pun turun. Komentar positif atau negatif bermunculan di twitter. Analisis sentimen digunakan untuk mengidentifikasi komentar positif dan negatif. Pada penelitian ini, komentar positif dan negatif akan diklasifikasikan menggunakan klasifikasi Vader dan nave bayes. Data yang terkumpul sebanyak 500 data dari twitter terkait down case facebook. Dari hasil perhitungan diperoleh sentimen positif sebanyak 33,92% dan sentimen negatif dengan hasil 66,08%. Berdasarkan hasil visualisasi dengan wordcloud, kata yang paling banyak muncul adalah kata facebook down untuk sentimen positif dan negatif. Hasil yang didapatkan dari tabel Confusion Matrix dari model klasifikasi menggunakan data sharing, 80% data training dan 20% data testing, dengan metode klasifikasi menggunakan Naive Bayes dengan pembobotan kata TF-IDF, nilai akurasinya sebesar 73,69% dan untuk Count Vektorizer adalah 70,18%. AbstractFacebook is the largest social media in the world. All social media applications made by Facebook cannot be accessed simultaneously in approximately 6 hours. This happens not only in Indonesia, but in all countries in the world, on October 5-6, Indonesian time. With this case, various comments and opinios from people on Twitter related to the Facebook case were down. Positive or negative comments popping up on twitter. Sentiment analysis is used to identify positive and negative comments. In this study, positive and negative comments will be classified using Vader and nave Bayes classification. The data collected was 500 data from twitter related to the Facebook down case. From the calculation results, positive sentiment was obtained as much as 33.92% and negative sentiment with 66.08% results. Based on the results of the visualization with wordcloud, the words that appear the most are the word facebook down for positive and negative sentiments. The results obtained from the confusion matrix table from the classification model using data sharing, 80% training data and 20% testing data, with the classification method using Naive Bayes with TF-IDF word weighting, the accuracy value is 73.69% and for the Count Vectorizer is 70.18%.

Copyrights © 2022






Journal Info

Abbrev

multitek

Publisher

Subject

Chemical Engineering, Chemistry & Bioengineering Chemistry Civil Engineering, Building, Construction & Architecture Computer Science & IT Control & Systems Engineering

Description

Multitek Indonesia : Jurnal Ilmiah is a journal published by the Technic Faculty, Universitas Muhammadiyah Ponorogo (Unmuh Ponorogo) in collaboration with Universitas Muhammadiyah Ponorogo Research and Community Service. Published twice a year (June and Desember), contains six to ten articles and ...