Natural language contextual matching advertising – Semantic matching ads

By Jason Menayan June 27th, 2008

SemanticsBack in January, I blogged on Proximic, a contextual ad network that used “word shape” to determine the context of a page, purportedly a better determinant for accurate ad matching than simple keyword-density algorithms (for example, it would do a better job of not showing ads for Mrs Field’s cookies on an article on how to clear cookies from your browser cache).

Today, Peer39, a contextual ad network, has announced its second round of funding ($8 million, led by Canaan Partners), and the release of is SemanticMatch product, which, similarly, aims to provide significantly better matching to the actual topical context of a page. The technology is clearly geared towards social media, where topical context is much more subtle, and where contextual advertising has been underperforming.

How does semantic matching work? Peer39 is vague about how their technology works (something that we at YieldBuild understand – when you have a really cool technology that works, you want to protect it), but it says that it targets the “intent and meaning” of pages rather than looking for preselected keywords.

Contextual matching, however, can look at the presence of clusters of words instead of the presence and density of any one targeted keyword. For instance, ads for antivirus software would look for not only the word virus, but spyware, computer, hacked, malware, etc, and while the presence of other words like inflammation, bacteria, immune, and sickness, would suggest an entirely different semantic grouping and presumably a different sort of ad.

Yahoo researchers have proposed syntactic-semantic matching in a research paper released last year, that looked at both syntactic (keyword) and semantic (categorical) ad matching vs a simple syntactic/keyword match. Their research demonstrated that the semantic component added relevance to ads, more strikingly when the page quality was high (i.e. enough natural language surrounding the topic was available).

Here are some other online advertising companies that tout semantic matching:

Update (6/30/08): Amiad Solomon, Peer39’s CEO, explains a bit how it works to TechCrunch’s Roi Carthy.

Bookmark and Share

This entry was posted on Friday, June 27th, 2008 at 11:07 am and is filed under Online Advertising. You can follow any responses to this entry through the RSS 2.0 feed. You can leave a response, or trackback from your own site.

One Response to “Natural language contextual matching advertising – Semantic matching ads”

  1. Natural language contextual matching advertising - Semantic matching ads « Blog and Buzz Says:

    [...] language contextual matching advertising – Semantic matching ads Natural language contextual matching advertising – Semantic matching ads Today, Peer39, a contextual ad network, has announced its second round of funding ($8 million, led [...]

Leave a Reply