There have been several efforts to model directly and naturally multidimensional databases. The focus of such works is to model cubes, hierarchies, dimensions, etc. In the paper of Cabibbo and Torlone a multidimensional database is modeled through the notions of dimensions and f-tables. Dimensions are constructed from hierarchies of dimension levels, whereas f-tables are repositories for the factual data. Data are characterized from a set of roll-up functions, mapping the instances of a dimension level to instances of another dimension level. It also discusses the logical modeling of multidimensional databases: the basic model remains practically the same, whereas ER modeling techniques are given for the conceptual modeling of the multidimensional database. A mapping to physical entities such as relations and multidimensional arrays is also provided. In other papers the authors have also presented a calculus for f-tables oriented to the formulation of aggregate queries as well as graphical query languages and an equivalent algebra is presented. I find that this work has been very influential for the area of multidimensional databases as it provides a clean and neat foundation.
a service of Schloss Dagstuhl - Leibniz Center for Informatics