Generating and Optimizing Graphical User Interfaces for Semantic Service Compositions

Semantic Web service composition is a discovery process in which a given set of requirements are fulfilled by dynamically locating and assembling semantically annotated services [5, 6]. Semantic annotation of Web services is a set of models that describe its properties (e.g., inputs, outputs, process), in a formal language such as OWL-S [2]. These models provide an unambiguous description of service properties by relating them to concepts belonging to Web ontologies. While dynamic service composition provides a flexible applications which can change according to service failures and other factors, it raises several questions regarding the way users interact with the generated applications. Specifically, it raises a challenge for usability, which is defined as the effectiveness, efficiency and satisfaction in which users perform tasks using a given system [1]. In traditional software development processes, the user interface is manually designed, implemented and tested in order to ensure its usability. In contrast, in dynamically composed applications, the functionality is not established during the design of the system. Therefore, the user interface cannot be designed, let alone tested, for usability. The conclusion is that the user interface should be generated dynamically as well, reflecting the temporal functionality of the application. The field of automatic generation of user interfaces attempts to formally define the elements of user interfaces, including presentation and interaction [4]. However, they do not deal with usability optimization as they presume the models are designed with usability in mind. Therefore, this approach will not suffice for dynamic compositions, as these compositions are not optimized for usability. The contribution of our work is in suggesting a method for optimizing the usability of dynamically composed applications, using formal methods.