DEEPVIDNET: A DEEP LEARNING AND IOT-DRIVEN FRAMEWORK FOR PREDICTING SHORT VIDEO POPULARITY

Abstract

With the exponential rise of short video platforms such as TikTok, YouTube Shorts, and Instagram Reels, predicting the popularity of short videos has become a key challenge for content creators and platform managers. Traditional metrics based solely on user interactions fail to capture the multi-dimensional aspects influencing video virality, such as timing, location, engagement type, and contextual cues. This paper proposes DeepVidNet, an intelligent framework integrating Internet of Things (IoT) data with deep learning architectures to accurately forecast the popularity trajectory of short videos. The proposed model captures heterogeneous data streams—user engagement signals, environmental factors, and device-level metrics— through IoT integration and processes them using Convolutional Neural Networks (CNNs) and Long ShortTerm Memory (LSTM) networks. Experimental results demonstrate that DeepVidNet outperforms existing popularity prediction models in terms of precision, recall, and F1-score, establishing a robust and scalable solution for real-time short video analytics