The collection and composition of relevant data sources, i.e., data discovery, is often the most difficult task in BI projects, and can benefit the most from collaboration. For example, a user might want to add a location dimension to a cube being designed. Building it from scratch, filling it with data, and specifying all relevant hierarchies is a daunting task, but it is quite likely that others have already accomplished (large) parts of it, and subsequently shared them. Several relevant location dimensions could be available; which one to pick depends on the best fit to your data (e.g., coverage of the whole world and not just Europe) in combination with good "star ratings" and additional ``ground truth'' provided by collaborative ontologies or even Wikipedia. Similarly, (definitions of) measures can be shared. Another traditionally hard task is data cleansing; here, procedures for specific data (such as de-duplication of names and places) can be shared and reused.

Main Advisor at Aalborg Universitet (AAU)
Co-advisor at Universitat Politècnica de Catalunya (UPC)