Spam Comment Detection Using the Ensemble Technique

YouTubeis a highly popular video streaming platform that allows users to share and spread videos covering a wide range of topics, including educational content and entertainment. Its vast collection of user-generated content makes it a popular platform for creators to showcase their storytelling skills and connect with a global audience. Additionally, YouTube’s interactive features, such as comments and likes, provide viewers with the opportunity to engage in discussions and express their appreciation for compelling narratives. However, YouTube’s interactive commenting feature has raised concerns due to spam comments, which can diminish the video’s impact and popularity, often containing irrelevant or promotional content. In this research, we applied feature extraction methods and developed machine learning models using YouTube datasets featuring PSY, Katy Perry, LMFAO, Eminem, and Shakira to detect spam comments. The ensemble of classifiers, including Random Forest (RF), Decision Trees (DT), and Support Vector Machines (SVM), yielded an impressive accuracy rate of $96.68\%$.