# Trending Videos: Measurement and Analysis

**Authors:** Iman Barjasteh, Ying Liu, Hayder Radha  
**arXiv:** [1409.7733](https://arxiv.org/abs/1409.7733)  
**Date:** 2014-09-26  
**Summary page:** https://www.summarizepaper.com/en/paper/1409.7733v1/

## Key points

- Trending videos capture viewers' attention over a short period and have the potential to become popular
- Lack of thorough study and analysis on YouTube trending videos
- Authors conducted comprehensive analysis by collecting and monitoring data from over 8,000 YouTube videos for nine months
- High-resolution time-series data on viewership and related statistics were analyzed to examine various aspects of trending videos throughout their lifecycle
- Investigation into profiles of users who upload trending videos to understand their role in getting content to trend
- Directional relationship analysis using Granger Causality (GC) revealed insights into viewership patterns of different categories of trending videos compared to traditional correlation measures
- Statistical attributes distinguish trending videos and channels from other non-trending YouTube content
- Highly asymmetric directional relationships among different categories of trending videos were discovered
- Distinct pattern of viewership towards popular categories while some categories tended to be more isolated

## Summary

In their paper titled "Trending Videos: Measurement and Analysis," authors Iman Barjasteh, Ying Liu, and Hayder Radha explore the phenomenon of YouTube trending videos. Unlike popular videos that have already gained high viewership numbers before being labeled as popular, trending videos are those that capture viewers' attention over a relatively short period and have the potential to become popular. Despite their significance and visibility, there has been a lack of thorough study and analysis on YouTube trending videos. To address this gap, the authors conducted a comprehensive analysis by collecting and monitoring data from over 8,000 YouTube videos for a duration of nine months. By analyzing high-resolution time-series data on viewership and related statistics, they were able to examine various key aspects of trending videos throughout their lifecycle. Importantly, since trending videos are declared as such just a few hours after being uploaded, the authors were able to analyze their time-series data across critical and sufficiently long durations. Additionally, the authors investigated the profiles of users who upload trending videos in order to understand the role these users play in getting their content to trend. They also conducted a directional-relationship analysis among all pairs of trending video time-series they monitored using Granger Causality (GC) with significance testing. This approach provided deeper insights into the viewership patterns of different categories of trending videos compared to traditional correlation measures. The results of their study revealed clear statistical attributes that distinguish both trending videos and their channels from other non-trending YouTube content. The authors also discovered highly asymmetric directional relationships among different categories of trending videos. Furthermore, their directionality analysis demonstrated a distinct pattern of viewership towards popular categories while some categories tended to be more isolated. Overall, this research contributes valuable insights into understanding YouTube's trending video phenomenon by providing an in-depth analysis based on extensive data collection and monitoring. The findings shed light on various aspects such as viewership patterns, user profiles, statistical attributes, and directional relationships among trending videos.