| NetSeer

What is a Concept?


It’s one of the top questions new customers and partners ask us, and rightly so. The dictionary defines a concept as “an abstract idea, or general notion.” Originating from the Latin word conceptum, concepts are tied to thought, frame of mind, and imagination. As it relates to ad targeting, we define a Concept (yes, capital ‘C’) as the cognitive associations between words and ideas that form naturally in your mind. Concepts derive meaning only as they relate to other entities.

In many ways, Concepts function like a round of word association. Quick: what’s the first thing that comes to mind when you hear or see the word “green”? It might be an ecological lifestyle. A color. Money. Or possibly the subject of recent legislation passed in Colorado. There’s no right or wrong. All of these different associations are simply Concepts in our book.

So, what of Rubin and Rorschach? It’s a matter of perception. The classic Rubin’s vase image can be perceived equally as a vase, or two facial profiles. The Rorschach inkblot is unlimited in what it holds, depending on the mindset of the viewer. Perceptions are different, and that’s an area where ad targeting gets tricky. Our collective perceptions, behaviors and experiences all factor in to the great ball of human intent. How this is manifested in advertising is another story.

The potential universe of “greenness”

We’re all familiar with common targeting approaches, such as keyword, semantic, contextual, and behavioral. Keywords effectively function as a literal, one-to-one match of input and output. Using our same example above, the word “green” must be present in the text of a webpage to be considered a match – albeit one rife with potential for linguistic ambiguity.

By contrast, semantic targeting is the domain of natural language processing and linguistic taxonomies. Categorizing “green” in the metaphorical hierarchy of kingdom and phylum might be clear, but things get murky down at the genus and species level. And yet accuracy makes the difference between capturing intent – or not. Trying to classify every linguistic use case of “green” is nearly impossible, and faces scalability issues: in many ways a semantic approach to advertising mirrors a site-and-section media buying approach more so than granular, page-level impression buying.

With Concepts, all the various ideas associated with “green” form naturally into discrete clusters. This enables the mere notion of green – expressed through surrounding ideas on a page – to come through loud and clear, without requiring the literal inclusion of the word itself. Looked at another way, you might associate “Las Vegas” with glitz or kitsch. A sunny desert holiday may come to mind, or an annual tech conference pilgrimage. The ability to accurately parse these clues and ferret out intent makes a material difference for advertisers and publishers.

While some of these ideas may seem commonsense, that’s because you’re human and it’s the way the mind works. But the idea of an ad-targeting algorithm understanding these same associations is pretty heady stuff. Using machines that emulate human thinking, it’s possible to understand that yesterday’s angry bird was fleetingly flappy in the collective consciousness. Having captured 53 million unique Concepts, and the relationships expressed between them in the NetSeer ConceptGraph, it looks like we’re on to something interesting, and different.

It just ain’t easy being green.