Grzegorz Lehmann

Bridging models and systems at runtime to build adaptive user interfaces

Adapting applications and user interfaces at runtime requires a deeper understanding of the underlying design. Models formalize this design, express the underlying concepts and make them accessible to machines. In our work we utilize runtime models to reflect the state of the interactive system (its UI respectively) and to change its underlying configuration. So called executable models combine design information, runtime state, and execution logic. From the perspective of adaptive UIs this allows the dynamic reconfiguration of UIs according to design information and the current state of the application at runtime. Dedicated elements of the model create a causal interconnection between model and user interface and facilitate a continuous information exchange between the two. This creates a feedback loop between model and UI where external stimulations influence the model execution and where projections to the outside allow the dynamic alteration of user interfaces.

Executable Models for Human-Computer Interaction

Model-based user interface development is grounded on the idea to utilize models at design time to derive user interfaces from the modeled information. There is however an increasing demand for user interfaces that adapt to the context of use at runtime. The shift from design time to runtime means, that different design decisions are postponed until runtime. Utilizing user interface models at runtime provides a possibility to utilize the same basis of information for these postponed decisions. The approach we are following goes even one step further. Instead of only postponing several design decisions, we aim at the utilization of stateful and executable models at runtime to completely express the user interaction and the user interface logic in a model-based way.

Meta-modeling runtime models

Runtime models enable the implementation of highly adaptive applications but also require a rethinking in the way we approach models. Metamodels of runtime models have to be supplemented with additional runtime concepts that have an impact on the way how runtime models are built and reflected in the underlying runtime architectures. The goal of this work is the generalization of concepts found in different approaches utilizing runtime models and the provision of a basis for their meta-modeling. After analyzing recent work dealing with runtime models, we present a meta-modeling process for runtime models. Based on a meta-metamodel it guides the creation of metamodels combining design time and runtime concepts.

A 3-layer architecture for smart environment models

Enriched with more and more intelligent devices modern homes rapidly transform into smart environments. Their growing capabilities enable the implementation of a new generation of ubiquitous applications, but also raise the complexity of the development. Developers of applications for smart environments must cope with a multitude of sensors, devices, users and thus contexts. We present a model-based approach for modeling of, reasoning about and controlling smart environments. A context model provides adaptive applications with a unified access to the smart home environment and, through a unique approach of utilizing executable models, also reflects its state at runtime. The presented approach supports runtime user interface adaption and reconfiguration for seamless interaction and has been successfully utilized to build several context-adaptive applications running in our smart home testbed.

Dynamic Distribution and Layouting of Model-Based User Interfaces in Smart Environments

The developments in computer technology in the last decade change the ways of computer utilization. The emerging smart environments make it possible to build ubiquitous applications that assist users during their everyday life, at any time, in any context. But the variety of contexts-of-use (user, platform and environment) makes the development of such ubiquitous applications for smart environments and especially its user interfaces a challenging and time-consuming task. We propose a model-based approach, which allows adapting the user interface at runtime to numerous (also unknown) contexts-of-use. Based on a user interface modelling language, defining the fundamentals and constraints of the user interface, a runtime architecture exploits the description to adapt the user interface to the current context-of-use. The architecture provides automatic distribution and layout algorithms for adapting the applications also to contexts unforeseen at design time. Designers do not specify predefined adaptations for each specific situation, but adaptation constraints and guidelines. Furthermore, users are provided with a meta user interface to influence the adaptations according to their needs. A smart home energy management system serves as running example to illustrate the approach.

Automated Usability Evaluation of Model-Based Adaptive User Interfaces for Users with Special and Specific Needs by Simulating User Interaction

Adaptive applications have the potential to help users with special and specific needs. However, evaluating the usability of such adaptive applications tends to become very complex. This chapter presents an integrated concept for the automated usability evaluation of model-based adaptive user interfaces. The approach is supposed to be used complementary to custom usability evaluations at an early stage of development. Interaction of a user is simulated and evaluated by combining a user model with user interface models from a model-based development framework, which is capable of providing different adaptation alternatives based on user attributes and the context of use. The main benefit of the approach is that no additional descriptions of the application’s UI and tasks need to be created for the usability evaluation because they are already available from the development process. As a result, different design alternatives and adaptation variants can be compared under equal usability evaluation criteria. Further, the complexity and costs for applying automated usability evaluation to adaptive user interfaces for users with special and specific needs can be reduced.

Development of context-adaptive applications on the basis of runtime user interface models

One of the challenges faced by developers of applications for smart environments is the diversity of contexts of use. Applications in smart environments must cope with continuously changing context of use, so the developers need to prepare them for a possibly broad range of situations. Since the developer has no access to all environments, in which her application will be executed, it must be possible to simulate different environments and evaluate the behavior of the application at design time. In our demonstration the designer has the possibility to simulate and modify a runtime context model and observe as her application adapts on the fly. In the underlying runtime architecture applications, defined as sets of models, are adapted automatically on the basis of the information held in the runtime context model. A visual tool enables the user interface developer to access and modify the models at any time and immediately observe the behavior of the application.

Utilizing Dynamic Executable Models for User Interface Development

In this demonstration we present the Multi Access Service Platform (MASP), a model-based runtime architecture for user interface development based on the idea of dynamic executable models. Such models are self-contained and complete as they contain the static structure, the dynamic state information as well as the execution logic. Utilizing dynamic executable models allows us to implement a rapid prototyping approach and provide mechanisms for the extension of the UI modeling language of the MASP.

Seamless Home Services

The growing number of smart devices providing services in peoples’ homes and the increasing number of services available via the Internet creates two separate worlds that have not been successfully integrated yet. To bridge these two worlds, we propose a Home Service Platform supporting the seamless integration of home-based and net-centric services in a smart home environment. A common Service Engine supporting inter-service communication and aggregation provides human accessible user interfaces via a Service Portal and allows representing devices through a Home Device Controller. The integration of the two worlds allows us to create a new type of service: Seamless Home Services.

Enhancing interaction with supplementary supportive user interfaces (UIs): meta-UIs, mega-UIs, extra-UIs, supra-UIs . . .

In order to improve the interaction control and intelligibility, end-user applications are supplemented with Supportive User Interfaces (SUI), like meta-UIs, mega-UIs, helping or configuration wizards. These additional UIs support the users by providing them with information about the available functionalities, the context of use, or the performed adaptations. Such UIs allow the user to supervise and modify an application interactive behavior according to her/his needs. Given the rising complexity of interactive systems, supportive UIs are highly desirable features. However, there is currently no common understanding of types and roles of supportive UIs. Enabling concepts and definitions underlying the engineering of such UIs are also missing. In order to fill this gap, the workshop seeks a discussion with a broad audience of researchers, who have experience with the design and development of supportive UIs.