Abstract
Nowadays, the volume of information handled by enterprises continues to increase. This information can be derived from their operational sources (databases) or also from their manipulated and exchanged documents. The decision-makers are interested by new applications in order to extract and deduce knowledge from information of documents. In this context, we propose a new multidimensional model for the OLAP (On-Line Analytical Processing) of documents, named CobWeb, based on a combination of different standard facets extracted from XML documents. In this paper, we describe the conceptual, logical and physical modeling of the CobWeb model.