Predicting popularity of online videos using Support Vector Regression

TL;DR

This paper introduces a Support Vector Regression approach using Gaussian RBFs to predict online video popularity, demonstrating improved accuracy and stability over existing methods by leveraging temporal, visual, and social metadata.

Contribution

The paper presents a novel SVR-based method for video popularity prediction that outperforms state-of-the-art techniques and incorporates social and visual metadata for enhanced accuracy.

Findings

SVR with Gaussian RBFs yields higher prediction stability.

Adding social and visual metadata improves prediction accuracy.

Method tested on datasets with over 14,000 videos from YouTube and Facebook.

Abstract

In this work, we propose a regression method to predict the popularity of an online video based on temporal and visual cues. Our method uses Support Vector Regression with Gaussian Radial Basis Functions. We show that modelling popularity patterns with this approach provides higher and more stable prediction results, mainly thanks to the non-linearity character of the proposed method as well as its resistance against overfitting. We compare our method with the state of the art on datasets containing over 14,000 videos from YouTube and Facebook. Furthermore, we show that results obtained relying only on the early distribution patterns, can be improved by adding social and visual metadata.