Youtube is one of the most visited websites in the world. There were 2.6 Billion visitors in 2021 and 2.3 Billion in 2020. The comments in the Youtube videos are visible to each visitor. Bad folks can use the comment section on their benefit to trick visitors to direct traffic to their website including dangerous malwares or even inproper content for kids.
Being a tech giant brings responsibility of controlling the content created by other users visible to public. The development team at Youtube was asked to sort out the bad/spam comments to avoid any side effects on the community, particulary to the kids. They need to correctly flag bad comments in Youtube videos without removing regular comments.
Data comes from the archives of UCI website. The comments were collected via the YouTube API from five of the ten most viewed videos on YouTube in the first half of 2015. All 5 are music videos including “Gangnam Style” by Psy, Katy Perry, LMFAO, Eminem, and Shakira.
The goal of this project to train a machine learning model that can predict whether a comment is a spam or not.
For this purpose, machine learning models were trained with Naive Bayes algorithms (Multinomial and Complement approaches). Both were sucessful at predicting whether a YouTube video comment is spam or not.