Suppose there is a large corpus of XML documents, each of which describes a movie released in the last 30 years (for example, extracted from IMDB). A movie enthusiast wants to make a list of interesting movies based on various criteria, such as, the genre, lead actors, directors, etc. She first decides to narrow the focus to just thrillers. However, she then has to look into each document individually, since only then is it possible for her to tell whether the combination of actors, directors, etc. interests her. This would be timeconsuming if the documents in question contain hundreds of tags each. Instead, she could use our tool Xoom (XMLZoom) which can extract and present the key information in every document. This would drastically cut down the time to go through the documents. She could then use Xoom to zoom into specific portions of each of the remaining documents, instead of opening and scanning them from top to bottom. In this proposal, we describe the construction of Xoom a...