Performance Comparison of Multiple Supervised Learning Algorithms for YouTube Exaggerated Bangla Titles Classification
Mirajul Islam, Nushrat Jahan Ria, Abu Kaisar Mohammad Masum, Jannatul Ferdous Ani · 2021
Exaggerated titles of YouTube videos are annoying. YouTube has been the most popular online video service for several years, with billions of subscribers and viewers. These videos have titles. To attract audiences, the YouTuber uses the most interesting words in the video's title. On YouTube, there are two types of titles: consistent and exaggerated. Usually, these exaggerated titles encourage people to watch videos but in reality, people don't get much entertainment by watching these exaggerated titles videos. To overcome this problem, we present a Bangla text classification analysis approach based on Natural Language Processing (NLP) on YouTube titles. We collect two different phases of video data on YouTube after watching so many videos. This classification assists in the discovery of YouTube videos with exaggerated titles. We used a total of six models in this research. Based on the results of research conducted, Convolutional Neural Network (CNN) has successfully classified the exaggerated Bangla title videos because it achieves results of 80%, 81.25%, and 76.47% for accuracy, precision, and F1 Score respectively.