A Meta-model for Diverse Data Sources in Business Intelligence

With the manifestation and evolution of Internet, several new types of data have emerged (videos, images, audio files, documents…). These new types of data classed as unstructured are more and more used and exchanged between IT systems, therefore their exploitation in Business Intelligence (BI) systems will absolutely provide a gold mine of information that guarantee a better and rich decision-making. Unfortunately BI systems don’t consider this sort of data and they are still limited to classical data sources: structured as Relational data source and semi-structured as XML files. Many research works separate the treatment and the design of a data warehouse that involves heterogeneous sources in order to avoid any problems of data integration and storage. However, the need for an approach that gathers diverse data sources still present. In this paper we appeal Model Driven Engineering (MDE) to propose a meta-model that assemble and describe all sort of structured, semi-structured and unstructured data sources such as relational, multidimensional, XML and NoSQL databases. Models conforming this meta-model will serve as an input for our BI process and for designing and modeling a data warehouse.