This very interesting paper attempts to transform the way advertisements are displayed on Web sites. Wikipedia matching, as this approach is called, attempts to match advertisements more accurately to the Web page’s content. To begin with, the authors assume that “if an advertisement is related to the Web page content, then it is relevant to the user’s interest.” In order to determine whether ads are related to the Web page, “[the authors] chose a set of Wikipedia articles evenly distributed by different topics to be the reference points.” Using the reference points, the ads that are similar in context to the Web page are found in the database, ranked, and displayed.

It would be interesting to see if the same technique can be reverse engineered to match Web pages with advertisements. Since advertising produces revenue, why not match the query with the advertisements, in order to produce the best-matching Web page?

Using Wikipedia, the authors present a groundbreaking concept where the ads are dynamically matched with the content on the Web page. This approach is very useful; since Wikipedia has a free online database of updated content, this method reduces the false positives from homonyms and words that have multiple meanings. The paper has a good flow, and it is a substantial contribution to the field of contextual advertising.

However, although the authors evaluate the Wikipedia matching solution using algebraic equations, they should have included a working example. As it stands, only a math major can truly appreciate the algebraic equations. Also, the paper fails to discuss the impact of multiple winners using the Borda method.