Work in this area has concentrated on the issue of complexity and expressibility of logical systems related to databases and Artificial Intelligence (Computational Logic), in particular nonmonotonic reasoning as well as fuzzy, many-valued, and modal logics. The mainstay of the theoretical work in the group is the study of structural complexity theory and finite model theory. The group is also currently involved in the analysis and implementation of logical databases (Datalog and disjunctive Datalog).
Work in this area currently focuses on efficient query engines for the most popular XML query languages, such as XPath and XQuery, as well as new and highly expressive languages for XML. We are currently developing what is already now the world's fastest XPath 1 engine, as well as a very efficient and scalable engine for XQuery on streaming XML data, FluXQuery. Moreover, we study theoretical aspects of query languages for semistructured data, such as their complexity and expressive power. A few of the group's other projects related to semistructured data and XML can be found here.
For enquiries contact Reinhard Pichler.
Work in this area elaborates on the visual approach to wrapper specification developed by the group and implemented by our startup, the Lixto company. Further information can be obtained through its Web site.
Research in knowledge-based diagnosis systems has in the last decade concentrated on the model-based approach, where the logic representation of a problem domain is combined with observations from a particular problem case to automatically derive a set of explanations that explain faulty behavior. The institute has over time been involved in a number of industrial cooperations, e.g., with Siemens Austria, dealing with the development of diagnosis systems for telecommunications equipment. Lately, research and application projects have dealt with the development of tools for the diagnosis of bugs in VLSI circuit designs, as well as model-based software debugging in general.
More generally, diagnosis and repair as well as qualitative modeling for diagnosis are investigated, including the integration of diagnosis and repair using planning in order to reduce the overall costs of the task, reasoning about temporal and spatial properties of diagnoses, and focusing techniques exploiting monitoring and purpose information. We investigate the relationship between qualitative reasoning about and model-based diagnosis of technical devices. Further research topics are the design of model-based diagnosis algorithms and modeling concepts as well as extensions of model-based diagnosis in abductive and consistency-based logic frameworks.
The goal of automated configuration is the composition of complex technical systems from a catalogue of parts. This is an application area of growing commercial importance with direct implications for knowledge representation research. A project funded by Siemens Austria resulted in the COCOS configuration tool which is now being applied at Siemens to the configuration of large and very large telecommunications systems. Research issues include the formalization of configuration as CSPs and the extension of existing CSP approaches to handle the specific requirements of the domain.
Intelligent scheduling is a very active research area of high practical relevance. In this area past research activities include the development of a system for interval-based representation of time-related information (TimEx), as well as a scheduling expert system for a high-grade steel plant applied by the Austrian B?hler-Kapfenberg steel plant. Recent activities include the study of the effectiveness of different scheduling algorithms, the representation of scheduling tasks as constraint satisfaction problems (CSPs) and the development of a reusable C++ class library for scheduling applications which is being used in a new cooperation with B?hler-Kapfenberg.
Current research in the theoretical aspects of scheduling deals with temporal, modal and fuzzy logics as well as graph grammars with emphasis on real-time planning and scheduling.
Real-world scheduling is decision making under vague constraints of different importance, often using uncertain data, where compromises between antagonistic criteria are allowed. Fuzzy scheduling has been investigated since 1979, mainly to handle uncertain temporal relations. We have described which constraints of a steelmaking scheduling problem would benefit from being fuzzified; fuzzy constraint relaxation techniques for general scheduling applications; how a certain, very critical duration can be adequatly predicted using fuzzy expert system techniques; how a new combination of fuzzy set based constraints and repair based heuristics that help to model real-world scheduling problems; how to evaluate the sensitivity to configuration changes; research issues and challenges in fuzzy scheduling.