The why and how of metadata
When you begin developing a metadata model for a digital asset management system, it can seem overwhelming. You begin thinking about all of the possibilities, and it seems limitless. The adage states that a picture is worth a thousand words, so if you have a thousand pictures, does that equal a million words?
Instead of thinking about it in those terms, perhaps it’s better to start small and look at a subset of your assets. Try to figure out what they represent and how expansive you will need the terminology to be. A key to this process might be asking yourself the question, “Why would I want to find these assets in the future?” Put it in full sentences. For example, you might have a video clip of a baseball game, so the idea behind searching for it might be “I’d like to see footage from the Cincinnati Reds 2010 season, in which Joey Votto hits a home run at Great America Park.”* Now you know you’ll need the team(s), the year, the player(s), the location, and the result of the play. That’s the beginning of a metadata schema right there.
As you work through your content and pull out additional pieces of information you would like to track, you can start to group similar pieces together. Obviously for sports footage there will be concepts that translate across most sports, such as the ones mentioned above. If your content is more diverse, then you will have much more to work through, but you will still likely find patterns in your content and you will start to see categories develop. This could be the beginning of a rudimentary taxonomy, in which you have top level concepts, and within those top levels you have sub-categories and further breakdowns of the ideas represented by your content.
What you will want to know is that your schema is flexible, and can accept additional concepts going forward. You will likely acquire or create new content that represents ideas not currently reflected in your assets. When that happens, you want the ability to insert that new concept into an existing taxonomy with a minimum of effort. Perhaps it’s an entirely new top level classification, or maybe it’s a sibling to one of your existing sub-categories. Wherever it fits in with your existing content, you will want to be able to alter the metadata schema accordingly.
This flexibility is what I think of as the modularity of a metadata model. That just means that you can easily insert new concepts and terms into the schema to reflect changes in your content. This modularity will also pay dividends on the retrieval end, as it allows your users to combine concepts from your schema in different ways, and thus target their searches more accurately. This concept is also referred to as faceted classification, or faceted taxonomy. Briefly, it just means that you classify your content in different, mutually exclusive, ways. A common example is seen on the Zappos website, which enables you to choose from Men’s or Women’s shoes, colors, prices, styles, etc. Those bits of information exist independently of one another, allowing you to combine them in an almost infinite number of combinations. If Zappos starts carrying a new shoe, they have an easy way to insert its descriptive terms into their taxonomy, even if it means adding an entirely new type of shoe.
Hopefully this helps as you begin planning your metadata model. It’s a challenge, but if you take your time and think about it up front, you’ll end up much happier down the road. Your metadata needs will likely change, and you can set yourself up for success by creating a schema that is flexible and that can adapt.
*Please note that I am a die-hard Cincinnati Reds fan, and have been for life.
