Data Analysis Systems
DAS conducts applied research in the following fields:
- Data management and analysis: e.g., cost efficiency for storage, transfer and searches with respect to time and storage space;
- Process analysis: e.g., identification of statistical correlations;
- Process optimization and control: e.g., virtual sensors for the chemical industry, energy efficient controls based on model-predictive approaches.
For this project definition, knowledge and model fragments of various types and different semantic levels often play a role, e.g., statistical correlations, prognosis models, rules and ontologies. The goal is to apply suitable methods from mathematics and computer science to make the available sources of information useful for a respective application. In addition to high performance computing aspects for parallelized and distributed data analysis and data mining, in particular the following technologies and approaches play a central role:
- Machine learning and knowledge extraction via semantic inference: e.g., interpretable models based on semantic technologies; semi-supervised and kernel-based methods and innovative similarity metrics; virtual sensors and model-based control approaches for optimum control of machines or improved prediction models
- Semantic data management: e.g., semantics, meta-modeling, data warehousing, stream data warehousing, knowledge representation models to improve data quality
- Stream data analysis: e.g., collection, handling and analysis of large quantities of data based on data warehousing, complex event processing, stream data mining and incremental learning.
Data management, knowledge representation, knowledge extraction from data, data-driven modeling for optimization and prognosis, model predictive control.