Configuration and Rating
Subproject B2 deals with the configuration and rating of services in dynamic contexts. Services are described by interfaces with functional and non-functional properties. In addition, there is knowledge, e.g. in form of rules, on how to configure services and which interfaces and properties such a composition can combine. of services. Subproject B2 deals with the configuration and rating of services in dynamic contexts. Services are described by interfaces with functional and non-functional properties.
In addition, there is knowledge, e.g. in form of rules, on how services can be combined and what interfaces and properties such a composition of services would have.
A configurator in his function as an OTF service provider should now respond to a request for a service with a composition with the desired property. For the realization, property, model and case based approaches can be used, which can make use of structural properties of the domains for the efficient solution of configuration problems. The core of the service specifications is the use of domain-specific knowledge.
If several configurations meet the requirements, the "best" configuration should be selected by an evaluation function. Typically, it provides a summary evaluation, which can consist of hard and soft evaluation criteria. The hard evaluation criteria include prices, resource consumption, etc. for which a measurement function can be specified. The soft criteria include, for example, the user's experience, which cannot usually be formulated precisely, such as knowledge of quality or search effort for similar cases. Within the scope of this subproject, evaluation functions are developed that take soft criteria based on the evaluation of application profiles into account when controlling the search and configuration. By automatically merging and splitting sub-valuation functions, the aggregated evaluation function should be able to adapt more quickly to changing environmental conditions (Optima), since it is not possible in large systems to make every change to the evaluation functions manually.