In order to access, visualize, aggregate and analyze their data, the organizations generally use the techniques of data warehouses and OLAP (On-Line Analytical Processing). We distinguish two categories of data: structured (such: databases), and unstructured (such: Documents). In order to store and manage these both kinds of data, organizations exploit increasingly the advantages of XML (eXtensible Markup Language); it constitutes an important source for decision-makers to help them to interpret and control the evolution of their business processes. In this paper, we propose a new multidimensional model of semi-structured data (XML documents), based on complementary dimensions (Content, Metadata, Structure, Semantics…) and we integrate also the concept of Tag Cloud for visualizing multidimensional tables, in order to help the decision-makers to better analyze textual data in the result of the OLAP queries.