Building a Metadata System
That's why Porter asserts that providing meaningful metadata is a difficult process. She says that although some automation is possible, creating effective metadata requires a lot of work and direct human intervention. According to Porter, "Creating metadata is time-consuming and you need people to do a lot of work. There are some things that certain products can do for you. For example, they may be able to tell you the aspect ratio of the image—whether it is horizontal or vertical, which can be done automatically on ingestion. But in the main, a lot of the metadata needs to be created by human beings."
At Getty Images, for example, Porter says they start the process with some information the photographers provide, usually high-level caption information to describe the significance of the image. This is particularly important, according to Porter, when what's going on in the image is not inherently obvious. The caption information gets placed in the Getty custom metadata tools (and that drives the keywording approach as well). "So while there are some things that the system can do to help you, what it clearly can't do is tell you what the picture is actually of. The only way to do that is with people," she says. After the photographer has a pass, the image goes to a professional library staff where more detailed metadata and keywords are applied.
Most companies have some sort of existing system of organization in place, even if this is just a folder system, but you can use this structure as a starting point when building a DAM. For instance, DAM vendor Extensis, maker of Portfolio, grabs information from the folder structure and makes them keywords. "In most cases," says Dan Harlacher, Portfolio product manager, "the information you want isn't embedded in a file. If you want to add things like keywords or description, it often isn't there." He says that when you add an asset to the database you can enter this information, but the software also extracts some information automatically based on the share name, the folder name, and the file name.
Bock points out that this type of existing structure can be applied in other ways such as the way a training department organizes curriculum. "If you are in a training department, you are producing a range of courses. Your courses have titles related to curriculum and some sort of course of study. So you immediately have an outline for metadata," he says.
Even with such existing systems there is still no free lunch when it comes to applying metadata, but you can set up a system intelligently so that the metadata gets pulled from the regular workflow as part of the process. Bock points out that "You don't want people to do extra work because they won't do it. You want to make sure they just do it naturally as part of doing their job." He offers an example of insurance company adjusters: A company may have a DAM to track adjuster car pictures. These pictures need to have metadata that includes the name of insured person, make and model of car, and claim number. Since the insurance adjusters routinely provide this information as part of their data-entry process, the company should design the system so that the DAM pulls in the data from the data-entry application. "The adjusters don't do anything differently because entering this information is what they are paid to do," Bock says.
"The most important pieces of any implementation," says Mark Titchener, solution specialist at ClearStory, "are to better understand how the customer is trying to operate as a business and represent and organize assets to help them translate that to components within the product itself to map directly to that." For example, the customer may have a complex folder structure, but combined with metadata and an organizational structure through collection, he says you can simplify that process as well as that of the end user. Because "ultimately," he says, "if an end user can find assets quickly, it's a more successful system."