Sentiment Analysis of Text Memes: A Comparison Among Supervised Machine Learning Methods

Abstract

Meme is a new form of content in social media. A meme contains sentiment towards a particular issue, product, person, or entity. Memes can be in the form of text, images, or images that contain text. Memes are entertaining, critical, sarcastic, and may even be political. Traditional sentiment analysis methods deal with text. This study compares the performance of four sentiment analysis methods when used on Indonesian meme in the form of text and images that contain text. Firstly, the extraction of text memes was carried out, followed by the classification of the extracted text memes using supervised machine learning methods, namely Naïve Bayes, Support Vector Machines, Decision Tree, and Convolutional Neural Networks. Based on the experimental results, sentiment analysis on meme text using the Naïve Bayes method produced the best results, with an accuracy of 65.4%.

Sistem Informasi Anotasi Data Penelitian : Pengolahan Data

“…Tujuan dari anotasi data adalah pelabelan dari suatu teks yang mudah dipahami oleh sistem. Penelitian yang digunakan dalam implementasi situs web anotasi data, yaitu penelitian Sentiment Analysis [6]. Penelitian ini melakukan anotasi data menggunakan data target positive, dan negative.…”

Sistem Deteksi Konten Negatif pada Teks Website Menggunakan Metode Random Forest

“…Namun, masih kurang efektif karena proses penerimaan aduan dari masyarakat dan instansi terkait, disesuaikan dengan dataset dari server pusat, sebelum akhirnya dilakukan pemblokiran pada website yang diadukan. Metode machine learning dan deep learning telah banyak diaplikasikan di dalam teks oleh peneliti [2]- [5].…”