Perbandingan Metode Naïve Bayes Classifier dan Support Vector Machine untuk Klasifikasi Cyber Harassment pada Twitter

Rezi, Iwardani Saputri (2021) Perbandingan Metode Naïve Bayes Classifier dan Support Vector Machine untuk Klasifikasi Cyber Harassment pada Twitter. Undergraduate Thesis thesis, Institut Teknologi Telkom Purwokerto.

	Text cover.pdf Download (420kB)
	Text Abstract.pdf Download (68kB)
	Text Abstrak.pdf Download (139kB)
	Text Bab I.pdf Download (220kB)
	Text Bab II.pdf Download (469kB)
	Text Bab III.pdf Download (378kB)
	Text Bab IV.pdf Restricted to Registered users only Download (750kB) | Request a copy
	Text Bab V.pdf Download (139kB)
	Text Daftar Pustaka.pdf Download (160kB)
	Text Lampiran.pdf Restricted to Registered users only Download (229kB) | Request a copy

Abstract

At this time technology has developed rapidly, more and more are interested in technology, especially social media. With the ease of being able to access social media, it is not uncommon for those who use it to commit crimes. Cyber harassment also known as online harassment can be in the form of threatening or harassing via email, instant message, social media or, posting information online. This case often occurs on social media, such as Twitter. For that, we need an appropriate classification method to solve the Cyber Harassment case from Twitter data, namely in the form of a collection of tweets obtained from crawling using API and Rapidminer. Tweet data consists of two classes, namely those containing harassment (class 1) and not containing harassment (class 0). The data is preprocessed to manage text or features. After that, feature selection is carried out, namely weighing the features with TF-IDF. This study uses the Python programming language and uses two methods, namely the Naïve Bayes Classifier and the Support Vector Machine to compare methods that have good accuracy and determine the performance of each method. The Naïve Bayes Classifier method uses the Complement Naïve Bayes model and the Support Vector Machine uses the Support Vector Classification (SVC) model. The results of the performance of each method by dividing the training data and testing data, namely 80%: 20%, indicate the Naïve Bayes Classifier method with an accuracy of 86.30%, precision of 84.51%, recall of 87.21% and, f1 score 85.84%. and Support Vector Machine with 89.56% accuracy, 83.62% precision, 94.5% recall and, f1 score 88.73%. Thus the Support Vector Machine method is better than the Naïve Bayes Classifier method and can be implemented for the Cyber Harassment case on Twitter. Keywords: Cyber Harassment, Twitter, Python, Naïve Bayes Classifier and, Support Vector Machine.

Actions (login required)

View Item