Dynamic multi-dimensional models for text warehouses

In this paper, we introduce a dynamic multidimensional model, which is suitable for building text warehouses. The dimensions are atomic semantic categories embedded in a familiar taxonomy. This approach to text warehouses requires a large number of dimensions, some of which may be not known in advance. Central to the dynamic multi-dimensional model is the meta-snow ake schema, which is a snow ake schema with an index table. The index table contains metadata on dimensions consisting of atomic and compound semantic categories. The documents stored in the warehouse are retrieved according to the semantic categories assigned to them. Such a text warehouse increases the precision and e ciency of document exploration.