By Saif Ahmed Adib & Aniket Mahanti

Social media has come a long way since its inception. From Facebook to TikTok today, social media has created a separate field of its own in the vast world of information technology.

Social media has come a long way since its inception. From Facebook to TikTok today, social media has created a separate field of its own in the vast world of information technology. Approximately 4.20 billion people use social media today. The rise in the use of social media can be credited to the increase in Internet penetration. Nearly 60% of the world had Internet connectivity in 2020.

Online video portals play a major role in this ever-growing world of social media and have been shown to make a considerable impact on both audiences and global reach. Today about 59% of the global Internet traffic is due to video streaming and this trend is expected to grow in the foreseeable future.

The domination of video portals has also been seen in the 2020 Global Internet Phenomena  Report from Sandvine, which tells us that ‘Video, Gaming and Social sharing’ comprise over 80% of all the Internet traffic. Among them, YouTube is the leader with over 15% of all traffic followed by Netflix at 11% in second place. According to the report, this rise is also due to the ongoing pandemic which is forcing consumers to rely on video portals as their primary source of entertainment.

YouTube is the world’s largest video portal for user-generated content (UGC), although many other video portals serve niche interests such as education (e.g., TED Talks) and adult entertainment (e.g., XVideos). While the video and channel characteristics of YouTube have been studied in depth by researchers[1][2], there has been limited analysis on niche video portals like TED Talks and XVideos. TED Talks video portal hosts prominent talks from influential speakers and currently has 3,700 videos. XVideos is a popular adult video-sharing site, which ranks among the top-10 most visited sites around the world according to analytics firm, Similarweb.

We conducted a characterisation study of thematic video portals with a focus on two diverging themes – education and adult entertainment. The reason for choosing differing portals is that we wanted to observe whether the theme of the portal played a role in its video or channel characteristics. We also wanted to explore how these video portals made use of their social networking features. We study the characteristics of videos for TED Talks representing the education theme. For the adult theme, we chose XVideos and concentrate on its channel characteristics.

Using active measurements, we collected meta-data on 6,721 video channels from XVideos and almost 2,700 TED Talks videos to discover that they have shown significant growth over the past decade. Channels from XVideos have been viewed more than 400 billion times in total while the TED Talks videos have been viewed almost 5 billion times. The average channel on XVideos has 22,928 subscribers. We also noticed significant growth in the number of channels being created on the sites over the years.

Another significant finding, observed on both TED Talks videos and XVideos channels was that the top 20% of most-viewed channels or videos were responsible for 75 to 90% of the total views, affirming the Pareto Principle [3] which researchers have observed in previous studies[4][5]. As a result, there is a ‘rich get richer’ effect where the top channels tend to remain longer on the front page. To break this monopoly, the recommendation system can be updated by the system designers and more flexibility of content discovery should be introduced to the viewers rather than them seeing the same videos or channels being repeatedly suggested on their main page. Brand new content by uploaders also needs to be consistently promoted.

We also found that viewers are not interested to interact with the social media features of video streaming websites whether it is mainstream or adult. For XVideos, most viewers tend to only rate positively as the average rating percentage is almost 98%. The same pattern was also observed in the case of TED Talks where the positive ratings for a video were much higher than the negative ratings.

The tagging system was another common characteristic between the two video portals. While TED Talks uses its tags to organise content in a curated manner, XVideos uses a free-form system where commonly used keywords are attached as tags to its channels. TED Talks also has a hierarchy in the structure of its tagging system with some tags being more dominant than others.

Other observations made for TED Talks include, videos being much longer and receiving the almost same number of comments as other traditional video portals[6]. Similar to tags and views, the top speakers and occupations tend to have the most videos. However, the representation of female speakers is low despite them getting almost the same number of average views as their male counterparts.

By analysing the key properties of videos and channels, our characterisation study aims to provide insights into these video portals and how they use their online social network of users. Irrespective of their themes, they all function as an ecosystem where uploaders and viewers rely on and interact with each other. The viewers need the operator to run the infrastructure and the uploaders to provide content. Similarly, the uploaders need the owners to manage their videos and the viewers to drive traffic into the portals.

References:

  1. Bartl, M.: Youtube channels, uploads and views: A statistical analysis of the past 10 years. Convergence: The International Journal of Research into New Media Technologies 24, 16-32 (2018). DOI 10.1177/1354856517736979.
  2. Shen, H., Chandler, H., Wang, H.: Toward efficient short-video sharing in the youtube social network. ACM Transactions on Internet Technology 18, 1-25 (2018). DOI 10.1145/3137569
  1. Mahanti, A., Carlsson, N., Arlitt, M., Williamson, C.: A tale of the tails: Powerlaws in internet measurements. IEEE Network 27, 59-64 (2013). DOI 10.1109/MNET.2013.6423193
  1. Wong, C., Song, Y.D., Mahanti, A.: Youtube of porn: longitudinal measurement, analysis, and characterization of a large porn streaming service. Social Network Analysis and Mining 10 (2020). DOI 10.1007/s13278-020-00661-8
  1. Yu, R., Christophersen, C., Song, Y.D., Mahanti, A.: Comparative analysis of adult video streaming services: Characteristics and workload. In: 2019 Network Traffic Measurement and Analysis Conference (TMA), pp. 49-56 (2019). DOI 10.23919/TMA.2019.8784688
  1. Siersdorfer, S., Chelaru, S., San Pedro, J., Altingovde, I. S., Nejdl, W.: Analyzing and mining comments and comment ratings on the social web. In: 2014 ACM Transactions on the Web 8, pp. 1–39(2014). doi: 10.1145/2628441.

Aniket Mahanti is a Senior Lecturer in Computer Science at the University of Auckland. His research interests are in the general area of computer networks and performance evaluation with an emphasis on Internet measurements.

Saif Ahmed Adib is a Mater of Science student at the University of Auckland. 

Disclaimer: The ideas expressed in this article reflect the author(s) views and not necessarily the views of The Big Q. 

You might also like:

Social media as the new wild west: How can we curb lawlessness on Facebook?

Why is it so hard to stop COVID-19 misinformation spreading on social media?