Discovery of Probable Dimensions & Facts From Raw Data Available in the File Systems and Oracle Relational Database Schemas for a Multidimensional Data-Warehousing Application
Multidimensional data-warehousing applications play prominent role in facilitating business intelligence, and subsequent strategic decision making for the industry. It provides a good starting point for data-mining applications, which cater to different segments of industry, academia, research and government sectors. This paper provides for discovery of probable dimensions and facts from typical data-stores of any organisation, be they file-systems or Oracle relational database schemas. In the case of file-systems, a preprocessing step is executed to port the data in the file-system into the Oracle relational database. The dimensions are then discovered based on the attribute data-types; relational constraints, wherever existing; based on domain-consistency, whereas, the facts are discovered based on the discovered dimensions and attribute data-types.