Big Data and Service Science

Selected Projects


In the project BigGIS, a new generation of GIS is developed and researched, which supports decisions in various scenarios, by means of new mechanisms and on the basis of large amounts of heterogeneous data, better and faster. The innovation lies in the comprehensive cooperation of time and space in the analysis, the consideration of significantly larger amounts of even unstructured and unreliable amounts of data and a continuous processing pipeline, which supports not only descriptive but also predictive and prescriptive as well as visual analyses. In the three application scenarios "Civil Protection", "Environment" and "Smart City and Health", prototypical problems are addressed and solutions are elaborated, empirically validated as well as further developed.

Contact: Dr. Tim Straub



The project CloUIT deals with the establishment of new IT sourcing models at an investment company in Frankfurt (Main). In the context of an already fully outsourced infrastructure and service landscape, the focus is set on the selection of new service offers and providers together with data privacy and security questions. The reliable selection and seamless integration of new services into existing infrastructures, with the option of changing the provider at any time, requires explicit knowledge acquaintance of the existing IT risk situation. This is obtained through the analysis of monitoring data with data mining methods and their aggregation.

Contact: Dr. Tim Straub

Data Quality Management in Corporate Financial Controlling

In the project Data Quality Management in Corporate Financial Controlling, the FZI develops, together with Bayer AG, analytical processes for defining early indicators for forecast accuracy in financial planning processes. Based on robust prediction techniques, metrics for planning quality as well as probabilistic networks and complex rule engines are derived from large amounts of historic planning (revision-) data in order to anticipate predictive accuracy and potential biases already when delivering or revising the predictions.

Contact: Prof. Dr. Thomas Setzer


Moderne Fahrzeuge werden serienmäßig mit zahlreichen Sensoren ausgestattet, durch welche der Betriebszustand und zu einem gewissen Grad auch das Umfeld der Fahrzeuge erfasst werden können. In einem Kooperationsprojekt soll geprüft werden, inwiefern eine weitergehende Auswertung dieser Routinemessungen mittels statistischer Auswertung Rückschlüsse auf die Beschaffenheit der Straßeninfrastruktur erlaubt. Das Gesamtvorhaben umfasst die sensorbasierte Merkmalserfassung, deren Vorverarbeitung, die Datengenerierung zur Funktionsabsicherung, die statistische Auswertung geo-temporaler Daten durch Methoden des statistischen Lernens sowie die Entwicklung der zugrundeliegenden Datenverarbeitungsinfrastruktur.

Contact: Stefan Otten