N-GRAMS ASSISTED YOUTUBE SPAM COMMENTDETECTION

Abstract

This paper proposes a novel technique for detecting intrusive comments or spam on the video sharing website - YouTube. We describe spam comments as those which have persuasive intent or those who deem to be contextually irrelevant for a given video. The prospects of monetization through advertising on popular social media channels over the years have attracted an increasingly large number of users. This has in turn led to the growth of the malicious users who have begun to build up automated bots, which are proficient of large scale orchestrated distribution of spam messages across multiple channels simultaneously. These intrusive comments damage the fame of a channel and also the experience of regular users. YouTube themselves have embarked upon this issue with some finite methods which blocks unsolicited comments that consists of links. Those methods have proven to be exceptionally unproductive as spammers have found strategies to bypass such heuristics. Standard machine learning classification algorithms are operative but there is always a possibility for better accuracy with novel methods. In this work, we aim to identify such comments by implementing conventional machine learning algorithms such as Random Forest, Support Vector Machine, and Naive Bayes along with certain custom heuristics such as Count Vectorizer which have proven to be very effective in detecting and subsequently combating spam commentary.