Imagine that we have just a free-form text search field as interface to a search engine, and that we can only parse the results it returns. What can we do to get an idea of what this search engine has to offer content-wise? This is the well known problem of acquiring a resource description when faced with uncooperative servers. We present a new approach to solving this problem, which is less bandwidth intensive compared to previous approaches, such as query-based document sampling.
Presented at University of Lugano (USI), September 7th 2009, Lugano, Switzerland.