Research Projects:

Extensibility in Object-Relational Databases
Modern object-relational database systems (ORDBMS) provide frameworks to logically extend the relational data model by complex data types and operators. A robust and efficient data management is essential for off-the-shelf ORDBMS to advance into non-standard application domains as Geographical Information Systems (GIS), Computer Aided Design (CAD), or Enterprise Resource Planning (ERP). The main focus of my research activities lies on the design of relational storage and access methods that enable commercial ORDBMS to deliver the scalability and performance required by many non-standard applications project homepage.

Database Integration for Virtual Engineering
The development, design, manufacturing and maintenance of modern engineering products is a very expensive and complex task. Today, thousands to millions of CAD files of a car or an airplane occupy terabytes of distributed secondary and tertiary storage. The main objective of this project is to develop techniques for effective and efficient management of huge amounts of spatial data. As main application we focus on spatial queries for digital mockup and on similarity search in large CAD databases. This work is realized and evaluated in close cooperation with the Volkswagen AG, Wolfsburg, and the Boeing Company, Seattle. Please find more details on our project homepage.

Using Data Compressors for Efficient Query Processing
The conservative approximation of spatial objects by a set of voxels results in high storage requirements. This leads to long query times, especially for high resolutions. Using data compression is a smart way to lower the resulting high I/O costs. Special care is taken to find a good compromise between I/O- and CPU costs. In addition, we analyze which degree of redundancy is most suitable for high resolution voxelized spatial objects project homepage.

Spatial Similarity Search in 3D Data
In modern application domains such as multimedia, molecular biology and medical imaging, similarity search in database systems is becoming an increasingly important task. Especially for CAD applications, suitable similarity models can help to reduce the cost of developing and producing new parts by maximizing the reuse of existing parts. Shorter product cycles and a greater diversity of models are becoming decisive competitive factors in the hard-fought automobile and plane market. We are working on effective and efficient similarity models for 3-D CAD data, which helps to find and group similar parts. project homepage.

We are developing an industrial prototype, called BOSS (Browsing OPTICS-Plots for Similarity Search). BOSS is based on the approach to evaluate similarity models using the hierarchical clustering algorithm OPTICS. BOSS is an interactive data browsing tool which depicts the reachability plot computed by OPTICS in a user friendly way together with appropriate representatives of the clusters. This intuitive illustration supports the user in his time-consuming task to find similar parts in large databases containing 3D CAD objects project homepage.

Further Research Interests
  • Distributed Data Mining
  • Parallel Data Mining
  • High Performance Data Mining
  • Spatio-Temporal Query Processing
  • Data Mining on Moving Objects
  • Geographic Information Systems