Query- vs. Crawling-based Classification of Searchable Web Databases
This paper describes a technique of classifying web based databases by pumping in combinations of terms into the web site's search box and looking at the distribution of numbers of results for each term combination.