Starry vault: Automating multidimensional modeling from data vaults

Abstract

The data vault model natively supports data and schema evolution, so it is often adopted to create operational data stores. However, it can hardly be directly used for OLAP querying. In this paper we propose an approach called Starry Vault for finding a multidimensional structure in data vaults. Starry Vault builds on the specific features of the data vault model to automate multidimensional modeling, and uses approximate functional dependencies to discover out of data the information necessary to infer the structure of multidimensional hierarchies. The manual intervention by the user is limited to some editing of the resulting multidimensional schemata, which makes the overall process simple and quick enough to be compatible with the situational analysis needs of a data scientist.