Abstract : In usual retrieval processes within large databases, the user formulates a first basic (broad) query to target and filter data and next, she starts browsing the answer looking for precise information. We then propose to perform an offline hierarchical grid-based clustering of the data set in order to quickly provide the user with concise, useful and structured answers as a starting point for an online exploration. Every single answer item describes a subset of the queried data in a user-friendly form using linguistic labels, that is to say it represents a concept that exists within the data. Moreover, answers of a given 'blind' query are nodes of a classification tree and every subtree rooted by an answer offers a 'guided tour' of a data subset to the user. Finally, an experimental study shows that our process is efficient in terms of computational time and achieves high quality clustering schemas of query results