Metadata: What You Need to Know (And Why You Need to Know It)

Of all the areas of expertise in the streaming media industry, three seem to rank particularly high in the pantheon of geekdom: encoding, server configuration, and metadata. But even among these, there’s a hierarchy of esoterica. Attempt to engage people in a conversation about encoding, and you’ll likely get a decent number of takers because we all touch content that we want to optimize at higher quality or lower bitrates. Bring up servers, which only the enterprise and CDN (content delivery network) geeks seem to focus on these days, and you’ll still get some positive response. But try to talk about metadata, and nearly everyone will show that pained look you get when pressed to dredge up an arcane tidbit from some irrelevant college class you didn’t want to take in the first place. "Uh, metadata. Of course. Sure. Listen, I have to go do this thing here."

But long after the industry has solved technical issues with codecs, servers, and CDNs, we’ll still be wrestling with metadata. Not just the plain old "data about data" metadata used by static content management systems that house mostly text documents, such as Word or PDFs but also temporal ("time-based") metadata used by audio and video content-creation and content-delivery systems. Far from being arcane or irrelevant, metadata is an essential part of today’s and tomorrow’s streaming media. You really do want to take this class.

Not yet convinced? Then let’s start with two reasons why you should care.

First, money. Simply put, media-centric metadata is big business. Take Gracenote, for example, a company best known for the database that drives iTunes’ automated album, track, and artist information. Gracenote started out in 1998 as an open database; it then morphed into a private enterprise model several years later. By 2008, when it was acquired by Sony for $260 million, its annual revenues were projected at about $40 million, all based on metadata acquisition and delivery. All Media Guide (AMG) has a similar story, having been acquired in 2007 by Macrovision for $82 million.

There are recent examples of investment in the space as well: Digitalsmiths, a Raleigh, N.C.-based data-mining company that works with major motion picture studios to create the metadata surrounding their prized assets.

Figure 1. Gracenote’s metadata drives iTunes, Yahoo! Music Engine, and more. It was acquiredby Sony in 2008 for $260 million.

Berry’s company has received significant angel and venture funding, including an oversubscribed round in 2008 and a more recent investment from Cisco.

Second, future growth. On stage at the most recent Streaming Media East show, Steve Mack pointed out to his panel’s audience that "metadata may sound really boring, but the things that can be done with metadata are very cool."

Mack, who wrote the bible on streaming media production and is now technical operations director for ScreenPlay, ought to know. His company uses time-based metadata to catalog, deliver, and analyze promotional video content, including movie trailers, music videos, and video game trailers. ScreenPlay started as a simple concept that its founder used to catalog videos for his video store and that has now expanded into metadata for an enviable entertainment client list. Clearly, metadata is here to stay and is attracting sizeable interest—and investment. So what do you need to know?

As you might imagine, metadata is a big topic; while this article will touch briefly on a few major factors, such as the types of metadata and the challenges that lay ahead, we’ll only just scratch the surface. For added insight, I’ve compiled a list of additional resources at www.workflowed.com/metadata.

In this article, we’ll cover the following topics:
• What is the difference between static and temporal (time-based) metadata?
• Why create temporal metadata?
• How is metadata created?
• What challenges does the industry face in creating and using metadata?