Pengaruh Pembobotan Social Network Analysis Terhadap Sentiment Analysis Untuk Subjek Orang

Rustanto, Ikhwan (2020) Pengaruh Pembobotan Social Network Analysis Terhadap Sentiment Analysis Untuk Subjek Orang. Masters thesis, Institut Teknologi Sepuluh Nopember.

[thumbnail of 05211550012002-Master_Thesis.pdf] Text
05211550012002-Master_Thesis.pdf - Other
Restricted to Repository staff only

Download (3MB) | Request a copy

Abstract

Berdasarkan data dari Kementrian Komunikasi dan Informatika (Kominfo) per Februari 2018, tercatat pengguna internet di Indonesia mencapai angka 142,36 juta. Angka ini menunjukkan peningkatan sebesar 54,68% dibandingkan dengan awal tahun 2016. Peningkatan pengguna internet di Indonesia ini menunjukkan era keterbukaan informasi yang semakin melebar. Keterbukaan informasi menyebabkan informasi terhadap kinerja pemerintahan semakin mudah diperoleh oleh seluruh lapisan masyarakat.
Hal ini mendorong penelitian tentang analisa sentimen terhadap kinerja pemerintahan, dalam hal ini adalah Pemerintah Provinsi Jawa Timur (Pemprov Jatim). Penelitian yang ada tentang analisa sentimen, umumnya hanya fokus pada kalimat atau postingan yang menjadi bahan analisa sentimen. Sehingga kurang sekali memperhatikan subjek atau orang yang melakukan posting. Dengan semakin maraknya penggunaan akun palsu maupun bot, membuat kredibelitas dari pembuat opini menjadi semakin diragukan. Dalam analisa media sosial terdapat metode untuk menganalisa keberadaan akun palsu ataupun bot, salah satunya social network analysis.
Berdasarkan hal tersebut, penelitian ini akan mengembangkan model kombinasi sentiment analysis dan social network analysis untuk mengetahui pengaruh pembobotan social network analysis terhadap analisa sentimen masyarakat pada Pemprov Jatim. Penelitian ini akan menggunakan nilai sentralitas dari social network analysis sebagai penambah bobot dalam sentiment analysis. Dalam hal ini, penggunaan algoritma Naïve Bayes akan dibandingkan dengan Support Vectore Machine untuk membagi opini masyarakat berdasarkan dataset dari Twitter dan Instagram Pemprov Jatim dan 5 media berita online yang memberitakan Pemprov Jatim. Hasil percobaan menunjukkan hasil berkebalikan dari yang diharapkan, metode kombinasi memberikan penurunan akurasi untuk kedua algoritma.
==================================================================================================
Based on data from the Ministry of Communication and Information (Communication and Information) as of February 2018, recorded internet users in Indonesia reached 142.36 million. This figure shows an increase of 54.68% compared to the beginning of 2016. The increase in internet users in Indonesia shows an era of information openness that is increasingly widened. Openness of information makes information on government performance more easily obtained by all levels of society.
This encourages research on sentiment analysis of government performance, in this case the Provincial Government of East Java (East Java Provincial Government). Existing research on sentiment analysis, generally only focus on the sentence or post that is the subject of sentiment analysis. So as not to pay attention to the subject or the person making the post. With the increasingly widespread use of fake accounts and bots, making the credibility of opinion makers becomes increasingly doubtful. In social media analysis there are methods to analyze the existence of fake or bot accounts, one of which is social network analysis.
Based on this, this study will develop a combination model of sentiment analysis and social network analysis to determine the effect of weighting social network analysis on community sentiment analysis in the East Java Provincial Government. This study will use the centrality of social network analysis as an added weight in sentiment analysis. In this case, the use of the Naïve Bayes algorithm will be compared with the Support Vectore Machine to share public opinion based on a dataset from Twitter and Instagram of the East Java Provincial Government and 5 online news media that preach the East Java Provincial Government. Experimental results show the opposite of the expected results, the combination method gives decreased accuracy for both algorithms.

Item Type: Thesis (Masters)
Uncontrolled Keywords: Sentiment analysis, Social network analysis, Naïve Bayes, Support Vector Machine ======================================================================================= Sentiment analysis, Social network analysis, Naïve Bayes, Support Vector Machine
Subjects: T Technology > T Technology (General) > T57.5 Data Processing
Divisions: Faculty of Intelligent Electrical and Informatics Technology (ELECTICS) > Information System > 59101-(S2) Master Thesis
Depositing User: Ikhwan Rustanto
Date Deposited: 31 Aug 2020 07:17
Last Modified: 26 Dec 2023 15:42
URI: http://repository.its.ac.id/id/eprint/81368

Actions (login required)

View Item View Item