DSV-IS'94 - Trip report

The workshop was held in the Monastery of the Holy Cross, run by monks of the Carmelite Order. It was attended by 33 delegates from 11 European and American countries. Most of the participants came from universities or public research institutes and some from industry. Twenty papers were selected for the final program from 32 submissions. Ten of them were presented, the other ten were considered in the discussion session of the workshop as its aim was to promote discussion rather than paper presentation.

The first two days were opened by an invited speaker : Jim Foley spoke about UIDE (User Interface Design Environment) an environment which starts from data models and pre/post conditions to automatically generate a user interface. These conditions allow the management of the large number of states in which a system can be. The environment supports a range of functionality (automatic generation of interaction objects, dynamic control of their enabling and disabling, generation of help, support of correct-preserving transformation). This tool was considered an interesting reference point for all the new tools which are being developed in this area.

Phil Barnard spoke about interactions with advanced graphical interface and the development of latent human knowledge. His approach draws upon a parallel model of the human information processing mechanism. A number of specific issues were considered in the blending of multimodal information and sensory information. Another very important issue was how formal methods might also be used for representing the properties of user cognition. The talk was illustrated with excellent animations of the model.

In the first part of the workshop, three sub-divisions were identified :

1. Modelling in Design of Interactive Systems. In this section the use of abstractions and models for supporting the work of software designers in the development process of Interactive Systems was considered. Different starting points for obtaining useful information for this purpose were analysed : visual programming, task analysis, conceptual design and so on. Modelling approaches using abstractions for basic interaction objects such as PAC or Interactors were used in several contexts. Another important theme was how to bridge the gap from the conceptual design of Interactive Systems to the determination of software implementation. The necessity of a structured approach to modelling was recognised. For this purpose two main approaches were identified : a task-centred approach where the system functionality is organised in such way as to reflect user tasks and object-oriented modelling.
   In abstract modelling and design an important issue is to connect components both at the conceptual and the architectural levels. Thus attention was focused on possible relationships and composition operators.
   Another important issue discussed was how to integrate task analysis and functional requirements. The model-based approach was emphasised and illustrated by, for example, an approach based on entity-relationship-attribute model (ERA) and on an activity chaining graph (ACG) resulting from the task analysis and an approach based on data-flow diagrams and attributed grammars to model the behaviour of the user interface.
2. Tasks and specification. This session was more oriented to introducing the user point of view in the modelling and design of user interfaces. Thus notations for specifying user tasks and for deriving further information from them were considered. In particular approaches for investigating properties about the relationships between the information presented by the system and that required by the user in order to perform some tasks were discussed. Even the problem of using abstractions as a means to hide details of tasks at various levels was considered, trying to identify which are the possible general abstractions levels which should be used.
   The importance of usability in the early phases of design rather than a post-hoc usability assessments was noted. This means comparing different design options in the various phases of the development process of an Interactive System. A particular design situation is defined by choosing an appropriate value for each design option among a set of possible values. Determining such a value heavily depends on the task description. Several usability requirements were considered (error tolerance, memory requirements, selection feedback,...). Furthermore, the problem of comparing and combining different modalities was discussed.
   It was recognised that whereas the enabling technologies for multimodal representation are growing rapidly, there is a lack of theoretical understanding of the principles which should be observed in mapping information for some task domain into presentations at the user interface. Even if such a mapping is tried, there is no method for guaranteeing the usability of the product in every design situation, though particular cases have already been investigated.
3. Formal specification. In this part of the workshop different notations and approaches to formal specifications of Interactive Systems were compared. By formal notation is meant a notation whose semantics have been mathematically defined. Both software engineering notations and specification notations for HCI were considered. A wide range of approaches were considered and discussed; starting with logic for the analysis of distributed actions, which is especially useful for verifying properties such as an absence of deadlock, predictability, reinitiability and availability of a command. Algebraic models of interaction which are at a level of abstraction and generality to the various functional models try to capture more explicitly essential characteristics of direct manipulation such as IO synchronisation and dependence. Specific applications such as interactive knowledge based systems were considered as case studies in which models and approaches previously developed were evaluated.

In the second part of the workshop participants were divided into three working groups. One group was on Users (end-users of systems). It had to consider issues such as how to include user perception in the specification of Interactive Systems and how this affected the design. Other problems to consider were general usability issues and role of user tasks.

The group on the Role of Formalisms considered problems such as neutral descriptions/freedom from notational bias; domain modelling; trade-off between cost of learning and pay-off for each notation (textual, visual); role of formal methods, notations in design, temporal modelling and reasoning. Here specific classes of applications were identified where the use of formal methods is particularly recommended. For example safety critical applications because in these cases the cost of a failure is very important with respect to the cost of the development.

Finally, the group on Role of Development Environments focused on: identifying the users of integrated developed environments; identifying necessary tools; a design model where software engineering and human-computer interaction concepts could be integrated; formalism in design environments; criteria for acceptability; and task modelling. Several types of designers were identified and for each of type, we identified the existing support and improved support which should be included in future generation tools.

Thanks are expressed to all the participants, particularly the working group reporters: David Duce (working group on Users), Michael Harrison (working group on Role of Formalisms), Noi Sukaviriya (working group on Role of Development Environments). Thanks are also due to the sponsors, Eurographics, ERCIM and CNUCE-CNR whose organisational and financial support were major factors in the success of the workshop. The workshop was considered by the participants to have been interesting and valuable and it as decided to hold an annual workshop on this topic in the future. The next will be held in June ‘95 in Toulouse (France) and there are already possible candidates for future years. The final results of the workshop will be published as a book.

F. Paternó
CNUCE-CNR, Pisa, Italy

(Reprinted from Computer Graphics forum, Vol. 13, No. 5, December 1994, pp. 365-366), Blackwell.