Keywords Don't Cut It

Data-Driven Thinking"Data-Driven Thinking" is a column written by members of the media community and containing fresh ideas on the digital revolution in media.

Today's column is written by Amiad Solomon, Founder & President at Peer39.

In an effort to improve the accuracy of its search engine, Google recently acquired Metaweb, a semantic Web company with a robust database of thousands of “entities” that the company claims are easier to match to search results than regular keywords, which can have multiple, confusable, meanings. Google’s hope is that this technology will help it return more relevant search results and increase the accuracy of targeted ads on its paid search platform by matching the best keywords with queries.

A quick view of Metaweb’s intro video reveals that the company believes it is too tricky and messy to try to understand text on the web; instead the company prefers to organize the web by “entities,” or titles for singular objects with one meaning. As innovative as this may seem, this platform is one of many that focuses only on isolated keywords, and ignores strings of text or phrases, which add a richer layer of meaning toward understanding text.

For the benefit of search, keyword targeting is an ample strategy since there is just the specific search term to find within a body of text. However, this solution falls well short of a real grasp of content, and in display advertising, keyword targeting is no replacement for true comprehension of what a piece of content is about. Semantics, or content-based targeting, extracts the core meaning of the page and not only specific keywords. A typical web page contains hundreds of keywords, but none of those may really represent the core content of the page. Additionally, while almost every keyword may have a text ad associated with it, display ads contain almost no text, and semantic analysis is needed to match the meaning of all the text of the page with the core message of the ad.

The display ad industry is becoming increasingly data driven. While keyword-based targeting is certainly a helpful data point to have, online advertisements bear the highest chance of success with a user when the ad message matches the meaning of an entire piece of content, in the way that a human processes it. People understand content as a long string of words that tie together and relate to each other, not as a series of disconnected keywords. In order for ads to be truly relevant, content, not keyword, clarity is what’s worth digging for.

Follow Peer39 (@SemanticizeMe) and AdExchanger.com (@adexchanger) on Twitter.

1 Comment

  1. Amiad, great post! I completely agree. At Sociocast, we look to understand the key subject (context) of a page using semantic/NLP techniques and Language Modeling. Of course, the theme or themes are typically composed of combinations of keyword "atoms" which are named entities. Great post!

    Reply

Add a comment

XHTML: You can use these tags: <a href="" title=""> <abbr title=""> <acronym title=""> <b> <blockquote cite=""> <cite> <code> <del datetime=""> <em> <i> <q cite=""> <s> <strike> <strong>