Techniques for logical design and ef fi cient querying of data warehouses

Sommario Logical design of data warehouses (DW) encompasses the sequence of steps which, given a core work-load, determine the logical scheme for the DW. A key step in logical design is view materialization. In this paper we propose an original approach to materialization in which the workload is characterized by the presence of complex queries represented by Nested Generalized Projection/Selection/Join expressions , in which sequences of aggregate operators may be applied to measures and selection predicates may be defined, at different granularities, on both dimensions and measures. Then, we propose a novel approach to estimate the cardinality of views based on a-priori information derived from the application domain. We face the problem by first computing satisfactory bounds for the cardinality, then by determining a good probabilistic estimate for it. The results we present here concern the computation of upper bounds for the cardinality of a view considering a set of cardinality constraints expressed on some other views. Finally, we deal with the problem of populating and refreshing the data warehouse, which typically involves queries spanning several tables over the reconciled schema. We present a structural method, based on the notion of hypertree decomposition, for solving these queries efficiently. Then, we extend this method in order to take into account also quantitative information on the data values.