Semantic Tools from hakia for On-line Advertising

At hakia, we have been developing our own Semantic Advertising System, which is at its final stages of testing. One of the branches of this development is the Contextual Advertising. We have developed a middle-ware system for contextual advertising which can be used by the 3rd party SEMs and on-line advertising warehouses.

To gauge the current state of things, we set up a demo system and experimented in comparison with the dominating player, Google Adsense’s test link. Several case studies later, we were surprised to find out that the Google’s ad targeting against the submitted content suffers seriously from poor relevancy significant percentage of the time. The on-line demo is available for interested parties, just send us an email and explain your interest.

The content below is about “BEAT GENERATION” but Google Adsense suggests ads like beat maker, Deadbeat father, beat DUI. On the same screen, hakia identifies the correct triggers (bottom-left), and Yahoo test ads in response to these triggers bring relevant ads (top-left).

While hakia consistently identified the relevant content in all test cases, Google’s poor performance puzzled us. If the performance of the Adsense test links is a good representation of the current state of contextual advertising, it would explain why most advertisers stay clear and why the CTRs are low in contextual advertising.

The cases we analyzed left no margin of debate for unknown factors like “user profiling”. For example, we tested a page talking about email spam at Department of Energy’s Web site. Google Adsense test links suggested ads related to Energy, whereas hakia correctly suggested email spamming. It is obvious that Google jumps on terms on the page using some sort of Term Detection, which is never comparable to real content detection based on semantic analysis.

Let me remind the reader of the fact that the actual problem in contextual advertising is the fast moving, dynamic content, but not the static pages where the content owner can somewhat optimize the ads manually one by one. The challenge is to do it accurately in an automated manner.

The middle-layer we have developed can be utilized in various ways. SEOs and SEMs handling millions of pages for keyword extraction, meta tag insertion, etc., can utilize hakia’s tool. Contextual advertising providers can deploy the hakia’s middle layer software to handle vast amount of dynamic pages literally “hands-free” with semantic precision.

To make real progress, we have to look beyond the obvious “backward looking” statistics which undermines the actual potential of contextual advertising due to the tainted practice in the past. If your Website’s traffic is coming 90% of time from search engines, then those people who are viewing your content have “search” in their mind. Relevant advertising next to your content serves their mind set, and maximizes your monetization capacity. Poor relevancy makes the ads look bad, not only at the onset of exposure but also on the entire advertising industry.

Let’s imagine the future where TV and PCs will merge. Semantic contextual advertising systems will be able to push relevant ads at the bottom of the screen as the conversations change from one subject to another in any TV program. Is the TV anchor saying something of interest? Right there you can click on a link to catch it and jump to a relevant Web page. That’s the future.

        

This entry was posted on Wednesday, November 12th, 2008 at 7:39 pm and is filed under News, Technology. 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.