Sports Asset Management & Storage Forum: Detailed Metadata Is Key to Useful Content
With increasingly massive libraries of content, the proliferation of multiplatform distribution outlets, and an ever growing stream of data and statistics, it has never been more difficult for leagues and sports networks to manage their valuable audio/video assets. For many, the key to retaining one’s sanity lies in the ultra-detailed metadata methodology and logging procedures deployed at ingest, making it easier to find, use, and distribute content.
“I think there is no such thing as too much data,” said Christy King, VP digital, technology R&D, UFC/Zuffa, during a metadata-focused session she moderated at SVG’s Sports Asset Management and Storage Forum last week. “There is just a better search result that can come from that data. We never stop adding [metadata] to our DAM [digital-asset management], but we focus on figuring out how to return search results that are meaningful. There is a middle ground that nobody has quite figured out yet between a rigid structured taxonomy metadata structure and a free-for-all of open fields, where people input whatever. Statistics, reporting, closed-captioning, audio — all of these pieces need to be brought together to create better methods of returning the most useful results.”
Automation Is Essential
Although detailed metadata is integral to a high-functioning media-asset-management (MAM) workflow, the glut of content in today’s video-heavy world makes automation of these functions equally important. Scalable automation platforms like Levels Beyond’s Reach Engine help content owners automate the tracking, storing, and accessing of digital assets through their production workflow.
“Many of our customers had just thought of metadata as file names, and it was very unstructured. That’s simply not very efficient,” said Levels Beyond President/COO Nick Rhodes. “The more you can automate the dropdown [menus], the nomenclature, and taxonomy so that it winds up being consistent, you’ll be all the better for it in the end.”
One approach to metadata automation that continues to gain in popularity is dialogue- and audio-based technology. Nexidia’s Enrich platform allows users to automatically convert dialogue to valuable metadata, without the need for transcribing or logging. By automatically tagging media with thousands of keywords and phrases, these organizations can create searchable metadata for easier, quicker search of relevant content.
“We don’t deal with spelling,” noted Chad Rounsavall, VP of sales, media, and entertainment, Nexidia. “If we misspell a driver’s name, we don’t care. As long as it is remotely close, we are going to find that clip based on a phonetic sound. It adds the ability to find things you might not have found the first time around because they were misspelled. But, at the same time,” he added, “we will return some false positives, so having the other metadata along with it helps sort that material as you go through it.”
Turner Sports Media Operations Manager Chris May hopes to see automation take his staff’s production to yet another level through the potential of analytics-based clip recommendations based on an editor’s history.
“We would like to see people create user profiles to catch analytics and user behavior. When you search on YouTube, you see, ‘If you like this clip, you may also like this.’ In our case, if a user sees a great LeBron James clip, we would love [the system] to suggest another, similar [clip]. Collecting those user metrics and determining who is searching what and how will allow you to [discover] more interrelated content.”
Of course, automation can get you only so far: the human element is required for the final creative decision.
“Automation does, at a certain point, have its limits,” said Warren Arenstein, SVP, business development, Primestream. “While analytic software like this is very helpful, there has to be a human editorial quotient at some point. People will have to look at the content and set criteria why it’s a special shot. As a former editor myself, there is a tendency to go with what is known, but, if you can get more quality [metadata] assigned to the content, that is going to be very helpful going forward.”
Metadata Consistency Across the Industry
Establishing consistent metadata approaches within an organization is a challenge in itself, but creating a uniform approach across several entities is an even larger undertaking. Leagues and networks constantly exchange assets, and rights issues complicate matters. As a result, many in the industry continue to call for a set of basic standards for metadata assignment.
“When our loggers [assign metadata], those code words are not universal,” said Allison Malan, associate director, media technologies, ESPN. “So, when we send media to Turner, for example, they don’t know exactly what is in it because they can’t understand our lingo. I think working with our partners better to develop a unified set of those terms would help everybody in the industry.”
However, the drastic differences between the necessary data for various sports and distribution outlets make a standard set of metadata conventions much more challenging.
“The specialty nature of everyone’s particular sport — the difference between two fighters in the ring and 30 cars going around the track — can become a little maddening,” said Rhodes. “But I do think that there needs to be a few bedrock dropdowns just to create a level of sanity. Many of our customers share rights to different leagues, and, in some cases, we see very inconsistent metadata and naming nomenclatures, [which] just make it twice as hard to actually find the content. So, within the marketplace, there is definitely a movement towards having a consistent nomenclature in order to find the assets you need more easily.”