A two-week lecture block on "Configuration and Diagnosis" held by Professor Kleine Büning will take place in the period from April 4 to April 19, 2013, in the lecture hall O2.
The lecture block is motivated mainly by the research topics of the SFB 901 and is offered specifically to master students.
The course "Configuration and Diagnosis" focuses on methods for modeling and solving analysis and synthesis tasks using knowledge-based techniques. In analysis tasks, particularly in diagnosis and supervision tasks, a classification of a system is done based on feature values (e.g. specifications or structural resp. behavioral descriptions). Assigned classes can be faults (e.g. characterized by abnormal feature values) or operating states. In that sense, analysis tasks are classification tasks that deal with (technical) systems existing in reality or at least in form of a model. In synthesis tasks, a system has to be created from predefined building blocks such that the resulting system exhibits specific characteristics.
The building blocks may be elementary blocks or subsystems abstracting sets of elementary blocks. Synthesis tasks subsume planning tasks (a plan is a sequence of actions to achieve a goal) and configuration tasks (e.g. building blocks as technical devices or software modules). In the latter context, selection problems (which building blocks), parameterization problems (which size of a building block), and aggregation problems (which position for a building block) have to be considered. This course presents a selection of modeling frameworks for configuration tasks and diagnosis tasks as well as adequate algorithms solving these tasks using knowledge-based techniques. Problem domains will be characterized according to the amount and type of knowledge at hand to solve given tasks. An example of such knowledge is the well-known "no function in structure" assumption for technical systems.
In particular, web service composition will be considered as an example configuration task.