MEBIS 2015 - Workshop on Managing Evolving Business Intelligence Systems (MEBIS 2015)

Chairs: Selma Khouri (National Engineering School for Mechanics and Aerotechnics (ISAE-ENSMA), France and National High School of Computer Science (ESI), Algeria), Robert Wrembel (Poznan University of Technology, Poland)
Website: http://www.cs.put.poznan.pl/rwrembel/MEBIS2015.htm...
Nowadays, a business intelligence (BI) technology is a worldwide accepted and obligatory component of information systems deployed in companies and institutions. From a technological point of view, BI includes a few layers. The first one is an ETL layer whose goal is to integrate data coming from heterogeneous, distributed, and autonomous data sources, which include operational databases and other storage systems. The integrated data are stored in a central database, called a data warehouse (DW), located in the second layer. Based on the central DW, smaller thematically-oriented data warehouses can be build. They are called data marts (DM). Data stored in a DW or DM are analyzed by multiple applications, located in the third layer.
An inherent feature of the data sources that fed the BI architecture with data is that in practice they evolve in time independently of the BI architecture. The evolution of the data sources can be characterized by content changes, i.e., insert/update/delete data, and schema changes, i.e., add/modify/drop a data structure or its property. Handling content changes at a DW can be implemented by means of: (1) temporal extensions, (2) materialized views, and (3) data versioning mechanisms.
The propagation of structural changes into the BI architecture is much more challenging as it requires modifications in all the layers in this architecture, forcing the layers to evolve. Three basic approaches to handling the evolution of data warehouses were proposed by research communities. These approaches are classified as: (1) schema evolution, (2) schema versioning, and (3) data warehouse versioning. Even though a decent contribution was made in this research area, there still exist multiple open research and technological problems, like querying heterogeneous DW versions, efficient indexing multiversion data, integrity constraints for mutliversion DW schemas, modeling multiversion DWs.
To the best of our knowledge, no solutions were proposed so far to support the evolution of the ETL layer and the analytical applications layer.
We observe that the research on evolving BI systems is still very active. Recently R. Kimball proposed the extension to its SCD concept, extending it with SCD type 4 to 7. The international research conferences and journals publish papers on evolving BI systems, e.g., DaWaK 2014, EDA 2014, EDA 2015, TIME 2014, Information Systems 2015.
Handling multiple and evolving states of some entities is a more general problem. An intensive research is conducted also in the areas of versioning XML documents and versioning ontologies. Last but not least, some NoSQL storage systems support versioning of data, e.g., HBase, Cassandra