Classification of Trending Videos in YouTube

Social media platforms are the most user-friendly platforms in multiple aspects such as social networking, content sharing, etc., at present. This paper studies the trending video content sharing on YouTube. Trending content has a high social impact and prediction whether a given content has a high trend or not, is timely. The main objective of this research is to develop a model that can predict whether a given content is trending or not. The major limitation in the current literature is that the authors are focused on following the direct implementation of trending content. This research specifically addresses this problem via a graph network embedded in the classification algorithm. The content authors can be considered as root nodes of a graph, and the published content can be considered as seed nodes. Both root nodes and seed nodes provide facts for high user engagement. This paper focuses on the binary classification problem of trending or not trending content in social media platforms. Before classification, dimensionality reduction is used to the extract principal components of the features of both the root node and the seed nodes. Eventually, the authors discovered that the logical regression outperforms in binary classification, and they were able to achieve 82.33% accuracy in the predictions in a novel prediction system. Furthermore, the authors have identified that the channels whose videos are trending on YouTube in Sri Lanka have more than one million viewers, more than 100,000 subscribers, and are mostly created by Sri Lankans.