What happens when you have more metadata for a database record than there are fields of data in the original record itself? If you’re like almost every other company in the world you don’t have to imagine this scenario. It’s a reality you’ve lived with for quite some time.
Every customer record in a CRM system, for instance, links in some form or fashion to data on every single interaction the customer has had with your company’s web site, Twitter feed, etc. That metadata about the customer’s behavior is massively larger than the “real” record in your database, which might already include every phone call and email interaction with the customer. In this case a “cookie” on a web site or an email address in DM campaign is the link to the vast reservoirs of behavioral data in associated metadata silos.
Behavioral action is not the only place we see the potential for generating vast amounts of metadata. Image files can also have all kinds of contextual information appended. This includes data on the people, products and objects in the image; thematic information; micro-second timestamps; source or author information; texture, color and more. There are several image search tools on the market— including an Austin start-up called Clarify—specifically diving in to this thorny arena.
If that isn’t intriguing enough, another recent example from MIT highlights how a short digital video file of an unmoving object can have valuable, deep metadata. MIT researchers used high-speed digital photography to document virtually static objects. (Sound waves and subtle air currents cause micromovements in objects that appear unmoving to humans.) The researchers gathered enough metadata on the micromovements of an object to accurately project its reaction to outside forces. The video clip (the data object) was appended with such an enormous amount of micrometre-level GPS data on the object’s imperceptible micromovements as to allow three-dimensional rendering of the object and projections on how the object would react to outside forces of varying strengths, intensities, and directions. This kind of metadata makes possible all kinds of predictive modeling about the behavior of inanimate objects while enabling holographic rendering and 3-dimensional printing.
So, for all the talk of central data repositories it seems that the sheer unfathomable volume of metadata generated means that we’re actually tending toward more and more silos of specific types of metadata that can be retrieved, unified, and analyzed quickly to inform future actions. What insights will we all see when we have our fingers on the pulse of all our customers and can use “lakes” of metadata to model future behavior of people, products, and organizations? This is hard to predict. We can say that we are likely entering a period where the “connected dots” between big data (the first two dimensions), micro-level GPS data (the third dimension), and predictive analytics (the fourth dimension) will lead us to some unexpected places in the not-too-distant future.