Whozat performs a semantic analysis of the material found on each person, and intelligently extracts a set of key words and expressions that cover at the same time the most important, varied and discriminative aspects of the people matching the query.

For example, with the query ("first name: Michael last name: Jordan ") you can see from the expressions that we are talking about an "american basketball player", even a "professional basketball player", who can "slam dunk", played for the "chicago bulls", has been featured in "sports illustrated", but also that you can "choose form over 300000 posters" about him, probably find out where he was "born", and there is a "statue" of him somewhere. However, we might have been looking for Prof. Michael Jordan, a specialist in "graphical models", or a Michael Jordan who works in a "camera and electrical department", or a "steak house" called Michael Jordan.

The expressions serve a double purpose:

• Provide an overview of the people relevant to the query, and of the different aspects of each one of them.
• Be an essential tool for the feedback system that allows to refine your results (see dedicated paragraph).