Pengaruh Akun Palsu Terhadap Analisis Sentimen Menggunakan Algoritma Support Vector Machine

Pratama, Rivanda Putra (2021) Pengaruh Akun Palsu Terhadap Analisis Sentimen Menggunakan Algoritma Support Vector Machine. Masters thesis, Institut Teknologi Sepuluh Nopember.

[img] Text
05211950012001-Master_Thesis.pdf - Accepted Version
Restricted to Repository staff only until 1 October 2023.

Download (4MB) | Request a copy

Abstract

Peningkatan pengguna internet di Indonesia menunjukkan era keterbukaan informasi yang semakin meluas, sehingga mempermudah masyarakat dalam memperoleh informasi melalui dunia maya. Hal ini menjadi salah satu pendorong pemerintah dalam menyebarkan informasi terkait COVID-19 melalui media daring yang ada. Selain sebagai media penyebaran informasi, media daring dapat digunakan untuk mendapatkan umpan balik dari masyarakat melalui media sosial. Berdasarkan umpan balik tersebut, pemerintah dapat mengukur tingkat kepuasan dari masyarakat terhadap kinerja pemerintah dalam menangani COVID-19 menggunakan teknik analisis sentimen. Penelitian mengenai teknik analisis sentimen sampai saat ini cukup mendapat perhatian. Namun, penelitian yang ada selama ini lebih berfokus kepada opini yang terdapat pada kalimat maupun komentar, serta kurang mempertimbangkan subjek akun yang melakukan posting. Di satu sisi, penggunaan akun palsu atau bot di media sosial menjadi semakin marak, sehingga kredibilitas dari pembuat opini menjadi berkurang. Berdasarkan permasalahan tersebut, penelitian ini akan mengembangkan metode untuk mengukur pengaruh akun palsu terhadap analisis sentimen. Algoritma klasifikasi sentimen yang akan digunakan adalah Support Vector Machine (SVM). Sedangkan, algoritma klasifikasi akun palsu diperoleh dengan membandingkan antara Naïve Bayes (NB), SVM, dan Deep Learning. Terdapat 12 fitur akun palsu yang digunakan pada penelitian ini. Data yang digunakan pada penelitian ini diambil dari akun Twitter milik Kementerian Kesehatan Republik Indonesia, Satgas Penanganan COVID-19, Dinas Kesehatan Provinsi Jawa Timur, dan Dinas Kesehatan Kota Surabaya. Hasil penelitian menunjukkan bahwa klasifikasi akun palsu menggunakan Deep Learning memiliki performa yang lebih baik dibandingkan dengan algoritma NB dan SVM. Selain itu, penelitian ini juga menemukan bahwa terdapat pengaruh dari akun palsu yang dapat menurunkan performa analisis sentimen, meskipun tidak signifikan. Penurunan performa analisis sentimen hanya sekitar 1,36%. Namun, visualisasi sentimen menunjukkan hasil yang berlawanan dengan dugaan awal penelitian. Akun palsu yang dibuat ternyata lebih menggiring ke arah sentimen positif, meskipun dari persentase tidak menunjukkan angka yang signifikan. Perubahan persentase pada sentimen positif hanya sekitar 5,05%, sentimen netral hanya sekitar 4,05%, serta sentimen negatif hanya sekitar 1%. ================================================================================================== The increase of internet users in Indonesia shows the era of free information that is increasingly widespread, making it easier for people to obtain information. This is one of the reasons for the government to disseminate information related to COVID-19 via the internet. In addition, the internet can also be used to get feedback from the public through social media. Based on the feedback, the government can measure the level of satisfaction from the community with the government's performance in dealing with COVID-19 using sentiment analysis techniques. Research related to sentiment analysis has been mostly carried out, but so far, it has focused more on opinions contained in sentences and comments and has not considered the subject of the account that posted it. On the other hand, the use of fake accounts or bots on social media is becoming more and more prevalent, so that the credibility of opinion makers is reduced. Based on these problems, this study will develop several methods to measure the influence of fake accounts on sentiment analysis. Sentiment classification uses the Support Vector Machine (SVM) algorithm. Meanwhile, the classification of fake accounts is obtained by comparing the Naïve Bayes (NB), SVM, and Deep Learning algorithms. There are 12 fake account features used in this research. The data used in this study were taken from the Twitter accounts of the Ministry of the Health Republic of Indonesia, COVID-19 Handling Task Force, East Java Provincial Health Office, and Surabaya City Health Office. The results showed that the classification of fake accounts using Deep Learning has better performance than the NB and SVM algorithms. In addition, this research also finds that there is an influence from fake accounts that can reduce the performance of sentiment analysis although it is not significant. The decrease of sentiment analysis performance is only around 1.36%. However, sentiment visualization shows results that are contrary to the initial hypothesis of the research. The fake accounts that were created turned out to be more likely to lead to positive sentiment, although the percentage did not show a significant number. The percentage change in positive sentiment is only around 5.05%, neutral sentiment is only around 4.05%, and negative sentiment is only around 1%.

Item Type: Thesis (Masters)
Uncontrolled Keywords: analisis sentimen, akun palsu, naïve bayes, support vector machine, deep learning, sentiment analysis, fake account
Subjects: Q Science > Q Science (General) > Q325.5 Machine learning.
Divisions: Faculty of Intelligent Electrical and Informatics Technology (ELECTICS) > Information System > 59101-(S2) Master Thesis
Depositing User: Rivanda Putra Pratama
Date Deposited: 20 Aug 2021 05:50
Last Modified: 20 Aug 2021 05:50
URI: https://repository.its.ac.id/id/eprint/88366

Actions (login required)

View Item View Item