YouTube: Spam Comments Filtration Using Hybrid Ensemble Machine Learning Models

Abstract

In today’s era most of the YouTuber’s are facing the major problem with electronic spam as troublesome Internet phenomenon. This work proposes a methodology for the detection of spam comments on the video-sharing website - YouTube. YouTube is running its own spam blocking system but continues to fail to block them properly. In this work, we examined several top- performance classification techniques for spam comment screening and proposed a novel methodology. In this work, we have analyzed such comments by applying conventional machine learning algorithms such as Naive Bayes, Random Forest, Support Vector Machine, Logistic regression, Decision Tree and will construct another model utilizing ensemble and hybrid approach. This paper proposed the YouTube spam comments detection framework, examined, and validated by using data collected from the YouTube using Naïve Bayes multinomial, Gradient Boosting, Random Forest and tested in Weka and Python data mining tools.

Hybrid deep learning model for YouTube spam comment detection

“…Researchers have extensively studied the detection of spam on YouTube. Classifying YouTube comments as spam and ham using machine learning [5]- [13], cascaded ensemble machine learning model [14], Markov decision process [15], artificial neural network [16], Microsoft structured query language server data mining tools [17], contextual feature based one-class classifier approach [18], hybrid ensemble machine learning models [19], multi-stage spam account [20]. Brain-inspired hyperdimensional computing [21], genetic algorithmic multi evaluation [22], n-gram assisted [23].…”