Building a YouTube Spam Filter with NLP

Created using ChatSlide

Group 3 (LMFAO) developed an NLP project focused on spam filtering in YouTube comments using the Youtube03-LMFAO dataset. Text feature extraction was applied, with a 75-25 train-test data split. A Multinomial Naive Bayes model achieved ~53.7% test accuracy despite high false positives. Findings emphasize challenges in spam detection, with recommendations for future improvements.