Exploiting Significance

I recently wrote about why our method of extracting the most important information from a page using semantic significance is the right approach for audience targeting. Now, let’s talk about what else we can use significance for.

User Insights

Semantic significance is a powerful tool to learn more about your users. An advertiser or publisher with a certain amount of first party data can upload the data into our database to view what significant terms characterize their user outside of what they already know. Sports fans, for example, who also happen to be avid gamers, small business owners who are highly interested in farming. By not destroying any information, but distilling it into a few statistically-significant terms, we quickly glean insights about our users.

Native Ad Placement

If a publisher were to feed a native ad into our platform, we can extract the significant terms from the ad itself. With this information, we have a number of options: we can use those terms to inform the pages where we display the ad, searching for all those pages whose significant terms match those of the native ad to an adjustable, quantifiable degree of similarity. We can also go beyond the contextual targeting of a native ad, and simply find all those users whose semantic profile contain the terms contained in the native ad (again, to an adjustable, quantifiable degree of similarity).

Brand Safety

We also use significance for brand safety. By actively searching individual pages for significant terms, we can dynamically ensure in real-time that a given page is an appropriate place for an ad. Relying on simply blacklisting URLs for brand safety is obviously a method fraught with imprecision: most advertisers do not want to blacklist The New York Times, but those same advertisers could also probably find more than a few pages in the Times where they would prefer their ads not appear. So, we can either manually designate a list of words or have the system to generate them by scanning the types of sites we would not want to appear on, and then the system will actively anti-target any sites that contain those terms.

Interested in reading more? Read about what Semasio and certain Quantum Physicists have in common.

Author: Jeff Adler

Jeff Adler is a Client Development Manager at Semasio, based in New York City.
Find me on: