Article Preview
TopIntroduction
Recent years have brought the tendency to develop mobile applications in human-centred iterations (ISO norm 13407) in order to increase the user-friendliness (Nielsen, 1994; Rogers, Sharp, & Preece, 2002) of the final product. Usually, this process includes a phase to (1) understand and specify the context of use, (2) specify the user and organisational requirements (3) produce design solutions, and (4) evaluate these design solutions against the requirements. With the continuous involvement of end-users in empiric user evaluations, these phases are iteratively executed until the intended application fulfils all requirements of the end-users (ISO norm 13407). All phases of this human-centred process can be very time-consuming and error-prone and thus they can increase the development costs and time. In particular, the third phase requires experienced programming skills since different kinds of prototypes need to be efficiently and effectively designed and implemented. Also, the fourth phase involves expert knowledge in usability engineering due to the fact that user evaluations need to be properly conducted and analysed in order to efficiently and effectively verify whether the requirements are met.
These are causes why Myers (1995) argues for the development and application of tools or toolkits in order to save time and money. Myers points out two main requirements for tools. Firstly, tools need to improve the result of a development process: the quality of the resulting product. Secondly, tools should also enhance the process itself: the ease of use and efficiency to run through the process.
In this paper we present the approach of all-in-one tools which support both the third and fourth phase of the human-centred design process. We call these tools user-centred prototyping (UCP) tools since they support the design and implementation of prototypes (third phase) as well as the prototypes’ evaluation and analysis (fourth phase) with end-users. As an example we introduce our UCP tool called MoPeDT (Pervasive Interface Development Toolkit for Mobile Phones) that supports the UCP of mobile applications where mobile phones are used as interaction device in a user’s everyday environment. This concept follows Alan Kay’s term Third Paradigm Computing and the concept of Pervasive or Ubiquitous Computing (Weiser, 1991) where users can either (1) directly interact with real world objects of the user’s everyday life or (2) mediated via an interaction device.
In the following, we illustrate typical steps of the process that we call user-centred prototyping – the third and fourth phase of the human-centred design process. These steps base on our practical experience (Rukzio, Leichtenstern, Callaghan, Schmidt, Holleis, & Chin, 2006). Then we describe existing tool-support for the UCP and typical tool features. The main part of the paper addresses MoPeDT and its validation via a feature and user study. Finally, we cover ideas to tool-supported execute empiric evaluations.
Design Specification of Prototypes
When producing a design solution, the application’s appearance and behaviour is typically first designed and later on implemented. For mobile applications, the design is often done by defining a model for the different screens and the application flow which typically resembles a state chart. In this model, each state represents a screen of the mobile application and from each of these screen states user interactions can call other screen states. Consequently, user interactions (e.g. the execution of a keyboard command) are represented by transitions in the model. With respect to the level of detail, the design specification can be modelled with low or high fidelity. Low-fidelity models are often specified with pen and paper since thereby the models can quickly and easily be changed. High-fidelity prototypes are often designed with the support of graphical tools (Rogers, Sharp, & Preece, 2002). Their look and feel is often quite similar to the final product. At early stages of the UCP, the design specification characteristically is less detailed than in later iterations. Figure 1 illustrates a high-fidelity design specification of a mobile application.
Figure 1. High-fidelity design specification of the appearance and behaviour of a mobile application