It’s time for the online video industry to start a collective discussion about how to implement a video ID system for user-generated content. Here’s what it should look like.

In today’s world of user-generated content, how do we find what we’re looking for? With so many millions of videos being uploaded to platforms such as YouTube, as well as multiple enterprise content management systems, how can we organize, store, share, and preserve all of that content?

There is a long history of metadata, tagging initiatives or consortiums for monetizable content. However, no non-proprietary, open, cross-platform systems have been developed for user-generated content (UGC). The reasons for the lack of a common UGC identifier system are both simple and complex. On the simple side, non-commercial UGC creators and owners are not incentivized to build and operate such a system as there is no apparent way to pay for it.

Conversely, the complex reason most likely lies with the operators of commercial organizations including YouTube, Yahoo, Facebook and Twitter, whom are very happy to keep their internal tagging systems proprietary. This practice keeps the UGC (content) that non-commercial users upload and share from easily migrating to other systems. e.g. YouTube does not give away a common identifier or meta-tag of a video on its system to Instagram or Vine. This insures that viewing of a YouTube UGC asset is viewed on YouTube, regardless of the social network that shared the link.

youtube blocked

Searching for a user-generated video is hard

Today we have the EIDR and IAB indexes for professionally produced premium content and advertising platforms, but this is neither applicable nor useful for millions of user-generated videos. Such videos are only gaining in popularity and significance in today’s modern society and they are monetizable! However, no viable system exists to normalize the categorization, tagging, and identification of millions of these UGC videos.

For example, I saw a music video I liked in South America last year. After leaving the country I started to search for the video, but found it was difficult without knowing the name of the artist, song or lyrics. I spent over a year searching for the video with no luck. I used everything I could remember about this video.

It was a Spanish artist, a Spanish song, indie genre, montage video, and five minutes long. I could explain the specific style elements of the video, like what happened in the scenes, what the opening monologue was about, what the overall feel was, but to no avail, the video was undiscoverable. It was not until a music video producer happened to show me the video again that I was able to rediscover it. After finding it, I was able to purchase the whole album. Yet, without the random chance of someone showing me the video, how long would it have taken before I finally found it again?

While this may seem like an inconsequential anecdote, it raises important issues of how we label and categorize our videos. How do we create standards that allow us to find the content we are looking for?

Creating metadata that matters for videos

To improve classification and discovery, we need to create a standardized identifier system for user-generated content – an intuitive and standardized system that should be implemented across all video platforms from YouTube to enterprise content management systems. This would benefit all users and creators of video content, advertisers, marketers, and improve analytics. The standard should:

  • Go beyond just title and artist, beyond date added, and other basic identifiers. When we watch a video, what do we remember about it? The name? Artist? Length? No. We remember the content. It’s key imagery, style and the mood. It is much easier for the average user to describe the video then remember the name.
  • Hook into third-party video platforms like YouTube, Vimeo, EQ Network, Brightcove and Ooyala, as well as enterprise systems such as SharePoint. Viewers would not only have a standardized and easy to navigate list to search from, but the system should collect data about viewer experience, and how we process and store information about the videos. With standard inputs, we create a common point to collect valuable information on user-generated content, and a more intuitive way to locate (or relocate) videos.
  • Generate metadata about the video automatically when a file is created and then automatically modified when uploaded. Watching or viewing then generates raw data that is captured as metrics to be aggregated and analyzed and then matched and paired with viewers for actionable analytics. Such a system must be accurate to be functional, and only with automatic metadata generation can accuracy be assured. Because of the daily volume of user-generated videos it would be impossible to regulate all of the content without technology that automatically generates and binds metadata to the video itself. The system should create unified tagging systems that allow the video to be discovered more easily, and would become the basis for building and categorizing large libraries of videos.

The benefits of such a standardized are undeniable. Not only would it increase ease of discoverability for viewers and increase the views for the creators, it could also lead to the monetization of user-generated videos. By having all UGC videos adhere to specific tagging standards, analytics would be improved so that advertising and marketing campaigns can directly target users based on the specific tagging standards. The need for such a system is obvious, and it’s time for the industry to start a collective discussion about how to implement a video ID system for user-generated content.

Sean Barger is CEO of Equilibrium.

  1. Huge problem here. We’re actually starting to solve at YourSports: http://www.yoursports.com/video/1555583

  2. There is also the problem of getting millions of users to use such a system. As far as I know search engines can’t tell what a video is about. They depend on user descriptions and tags. The user would have to be involved in any more complex system of video identification unless the video platforms want to hire hundreds of thousands of workers to watch every uploaded video and insert the appropriate data.

  3. Geotagging would put the videos in context, for example :
    were shot from this location : “38.418967,141.3388″
    Optional search parameters such as distance , date and other tags ,
    would show only relevant content in or around a particular area
    (in this case Japan Tsunami)
    (PS: did someone notice a woman drowning in the 2nd video ?)

  4. Brian PlackisCheng Monday, September 23, 2013

    Great article. Media should be as searchable as web pages. At cielo24, we are turning media into a searchable database of words with time stamps that can be used for indexes, transcripts and captions (searchable captions). The barriers to creating high quality searchable captions are going away. The cost is nominal and the process is fast and easy. The return on investment is very high in terms of expanded audience and content discovery. The industry needs to rally to create searchable meta data standards. Happy to discuss offline how we can push forward.

    BrianPC, Founder and CEO
    cielo24 – searchable captions


Comments have been disabled for this post