Adaptive and Dynamic Quality-Aware Service Selection

The need for replacing services belonging to a composition is motivated by several reasons, such as changes in the application´s requirements, bug fixing, existence of a fresh service and so on. Due to the large number of services having similar (or even identical) functionalities, it has been widely accepted that the selection process for a new service should also take into account non-functional requirements (QoS attributes), such as performance, availability, security and so on. Existing approaches for service selection are usually static and do not consider quality attributes, i.e., they adopt a strategy (ranking algorithm) to rank the candidate services that is usually based on functional aspects and is never altered. In this context, this paper proposes a solution that allows to change the ranking strategy at runtime based on historical data of quality attributes.