XML archive for testing : a benchmark for GuessXQ
Résumé
Aiming at making easier the process of information retrieval from structured documents, we developed an environment, GuessXQ, to provide query-by-example assistance. GuessXQ offers the user the chance to choose a family of documents, and to annotate graphically a sample document, picked from the collection. From the visual annotation, that exemplifies the user's needs, GuessXQ infers a XQuery query that automatically applies to the collection to retrieve all the documents satisfying the information required by the user. However, after developing a GuessXQ prototype, it is mandatory to validate pragmatically the effectiveness of the environment. This can be done, drawing a set of experiments. The success of such assessment strongly relies on the quality of collections used. This claim led to the construction of XAT, an archive of XML collections specially gathered in order to be useful for testing XML tools. Looking for a generic purpose, XAT was organized according to a classification criteria developed for this specific purpose. This article is about the design and implementation of XAT.