Gregor Buchholz

Task Model-Based Usability Evaluation for Smart Environments

Task models are widely used within the research field of HCI for the model-based development of interactive systems. Recently introduced approaches applied task models further to model the cooperative behavior of people using devices within a smart environment. We describe a method of model-based usability evaluation to evaluate interactive systems, with a particular focus on smart environments, which are developed based on task models. We consider the evaluation in early development stages to interactively walk through the models and in later stages to execute a test case within a real environment. The paper provides results of a prototypical implementation.

Integration of usability evaluation and model-based software development

Model-based software development is carried out as a well defined process. Depending on the applied approach, different phases can be distinguished, e.g. requirements specification, design, prototyping, implementation and usability evaluation. During this iterative process manifold artifacts are developed and modified, including, e.g. models, source code and usability evaluation data. CASE tools support the development stages well, but lack a seamless integration of usability evaluation methods. We aim at bridging the gap between development and usability evaluation, through enabling the cooperative use of artifacts with the particular tools. As a result of integration usability experts save time to prepare an evaluation and evaluation results can be easier incorporated back into the development process. We show exemplary our work on enhancing the Eclipse framework to support usability evaluation for task model-based software development.

Subject-Oriented Specification of Smart Environments

In this paper, we describe a subject-oriented approach for specifying possible activities in smart environments. The approach is based on task models for subjects and a team model for representing the results of the cooperation. This model can be considered as communication model in the sense of S-BPM. Tasks in the team model have triggers for starting and finishing them. The triggers are the result of task executions in instances of models of subjects. A very restricted notation of OCL is used for those purposes. The same kind of simple OCL expressions can be used to specify preconditions of tasks. Tasks in the team model cannot be executed directly. They only present the result of other models. However, team models can also restrict the execution of other models. Additionally, context definitions are introduced for task models. Within a certain context, variables can be bound to specific values. This allows a very detailed description of activities that is more precise than most workflow specifications. Communication can be restricted to certain instances of subjects.

Extended Features of Task Models for Specifying Cooperative Activities

In this paper, an extended version of task models is discussed that allows detailed specifications of cooperative activities. Within the presented specification CoTaL (Cooperative Task Language), there exist two complementary types of task models called role model and team model. One or more instances of each role model describe the specific activities of actors. The team model represents joined activities and reflects progression in cooperation between role instances. For each scenario there exists one instance of the team model. Preconditions and events can be assigned to tasks and refer to one or all running instances of a role model. An event can be a starting or finishing trigger and is activated as result of task executions. Additionally, variables can be defined. They are bound within the specified context during runtime and get the value of the identifier of a certain role instance. In this way, communication and collaboration between different actors can be specified. Tasks of a team model cannot be performed directly but present the result of the execution of other (role) models only. However, a team model can restrict the execution of role model instances. It is shown how such models can be used to specify the activities in a smart meeting room. Snapshots of their simulation in CoTaSE (Cooperative Task Specification Environment) are presented. Additionally to the local implementation there exists an implementation in a cloud. It allows real cooperative executions of tasks.

From Collaborative Scenario Recording to Smart Room Assistance Models

There is still much to be done in implementing assistance systems for Intelligent Environments. Different approaches exist that aim at providing the user with useful and pleasant functionality. One group of methods uses behavioral models to derive supportive actions from the observation by sensors. This is a promising approach but creating such models is a laborious and error-prone task. Examples of the behavior of persons in intelligent environments and their interactions with the devices are a starting point for the partial generation of such models. In this paper we present an approach to record user behavior without the need of real users performing in the real environment. As a special thematic priority we will focus on the preparation phase of collaborative scenario recording and the used notation. Additionally, the paper will explain the generation of models from the recorded traces.

Models as a Starting Point of Software Development for Smart Environments

Creating a smart environment is a challenging task because of the excessive software development and adaptation required. Additionally, hardware in form of stationary as well as dynamic devices has to be installed. Similar to traditional software development, evaluating only the end product is often very costly in terms of time and effort needed. This is due to the fact that usually a lot of changes have to take place since the system fails to deliver the expected behaviour. Therefore, modelling is of great benefit. Models help to get a shared and thorough understanding of a specific domain. Making the animation of those models feasible allows getting a first impression of the system under development. Such prototypes of a system can be created on different levels of abstraction. The paper aims to demonstrate how modelling the human behaviour from the perspective of the activities performed in the environment can lead to first abstract prototypes. Those prototypes can be further extended and fostered by device models as well as models for the whole environment. In the paper, we also strive to discuss the costs and benefits of offering an abstract environmental model in 2D or 3D. 3

Managing Complexity in Activity Specifications by Separation of Concerns and Reusability

The specification of activities of the different stakeholders is an important activity for software development. Currently, a lot of specification languages like task models, activity diagrams, state charts, and business specifications are used to document the results of the analysis of the domain in most projects. The paper discusses the aspect of reusability by considering generic submodels. This approach increases the quality of models. Additionally, the separation of concerns of cooperation and individual work by subject-oriented specifications is discussed. It will be demonstrated how task models can be used to support subject-oriented specification by so called team models and role models in a more precise way than S-BPM specifications. More precise restrictions on instances of roles can be specified.

Elaborating analysis models with tool support

Integrating models as essential elements into the software development process is supported by numerous methods and tools but the creation of such models still bears a considerable challenge. This paper proposes a structured modeling of tasks and activities during the requirements analysis in order to pave the way for the very early utilization of models. A tool implementation demonstrates the elaboration of models based on scenarios.

Integrating Usability Methods into Model-Based Software Development

Model-based software development is carried out as a well-defined process. Depending on the applied approach, different phases can be distinguished, for example, requirements specification, design, prototyping, implementation, and usability evaluation. During this iterative process, manifold artifacts are developed and modified, including models, source code, and usability evaluation data. CASE tools support the development stages well, but lack a seamless integration of usability evaluation methods. We aim at bridging the gap between development and usability, through enabling the cooperative use of artifacts with the particular tools. As a result of integration usability, experts save time to prepare an evaluation and evaluation results can be easily incorporated back into the development process. We show exemplary our work on enhancing the Eclipse framework to support usability evaluation for task model-based software development.

Towards Usability Evaluation for Smart Appliance Ensembles

Smart environments comprise users and devices to form ad-hoc an ensemble and assist the users to fulfill their tasks more efficiently and more conveniently. This introduces new challenges for usability evaluations. To cope with theses issues, we propose the application of task models. Following this approach the behavior of the users can be interpreted as a trace through the corresponding task model. We discuss our method of capturing, visualizing and analyzing traces through task models within smart environments. The paper provides the first results of a prototypical implementation.