Automated Generation of User Interfaces - Based on Use Case or Interaction Design Specifications?

Instead of manually creating (graphical) user interfaces (UIs), automatically generating them is desirable, especially since UIs are needed today for diverse devices like PCs and smartphones. The basis for such automated generation can be a UI specification, but most of the related research takes task models as input, which are on a higher level of abstraction. More recently, another modeling language employing discourse-based models for specifying communicative interaction has been proposed, which completely abstracts from specifics of a particular UI and even its type. From such models, UIs can be generated automatically through model transformations. Some research, however, claims that UIs can be generated from use cases. While it would be desirable to utilize such a form of requirements definition without having to create another specification, we found that these approaches actually use additional information attached to the use cases, usually UI-related attachments. In addition to contrasting different kinds of specifications, we propose a synthesis through a combination. In fact, we found that discourse-based models can be also viewed as specifying classes of scenarios, i.e., use cases more precisely than the main-stream approach using UML and use-case reports.