ECCE - 2 - Second European Conference on Cognitive

MODELS AND METHODOLOGY Representational Frameworks And Models For Human-Computer Interfaces Gabriele Rohr, Michael J. Tauber IBM Germany Heidelberg Scientific Center, Federal Republic of Germany This paper is mainly based on discussions between a computer scientist and a psychologist concerning models of human-computer interaction. It tries to specify what a model is, which purpose it serves, and which components of human-computer interaction have to be modelled. Furthermore, these specifications are compared with already existing models. Models are discussed recently in connection with building up an adequate user interface architecture. Design criteria are needed to construct interfaces which take into consideration human information processing abilities as well as task structures represented by the human. Hereby, "Architecture means the complete and formal description of the surface of a system seen from a well-defined interface. Therefore, architecture is more than the usual specification. Architecture also contains a model of the user and a model of the communication between a user and a system ... Architecture does not refer to the product only. With the same weight, architecture refers to the production process and its documentation" (Zemanek 1982; translated by the authors). To meet these requirements, models of human behaviour in interaction with computer systems are needed. Several models of human-computer interaction have been worked out in the past. They differ however very much in the aspects they describe. A classification of these models has not been done yet. In this paper an attempt is made to clarify the knowledge and methods that are required to build up an adequate model of human-computer interaction and would help to formulate an abstract architecture. Central roles in the discussion play the terms model and representational framework. It must be pointed out that the first step in building up an architecture is the complete formal specification of the virtual system from an intended user's point of view. On Models And Modelling In Human-Computer Co-Operation Horst Oberquelle Universität Hamburg, Fachbereich Informatik, Schlüterstr.70, D-2000 Hamburg 13, Germany "System designers take note. Design the system for the person, not for the computer, not even for yourself .... Provide the user with an explicit model." D. A. Norman (Norman, 1981) In the recent literature on human-computer communication (HCC) 'models' have become a central notion. State/transition models of interactive systems are discussed again and again (from Parnas (1969) to Jacob (1983)), the sites/modes/trails model of Nievergelt is praised as a solution for problems many users encounter (Nievergelt, 1983), the model human processor is designed to analyse user behaviour (Card, Moran & Newell, 1983), an abstract model of communication between two partners has been applied to HCC (Oberquelle, Kupka & Maass, 1983). There are lots of other papers discussing conceptual models or mental models of interactive systems, user models or the role of metaphors (see e.g. Carroll & Mack, 1982). Obviously, humans need and use models of their 'relevant system' to be able to plan their actions and to control the obtained effects. As soon as several persons cooperate their communication will be based on their respective models. Part of their communication will serve to develop, to explain and understand models. To communicate about models presupposes that models can be represented externally and that the means of representation are known to the communicating parties. Communication about modelling tools might be necessary in addition. All kinds of modelling, formal models included, are based on the human ability to communicate in the informal mode inherent in natural language (cf. Naur, 1982). Users of computerized systems have great difficulties in acquiring adequate models of the systems they are forced to work with. The development of systems for HCC is not just the creation of programs according to some specification, but the design of working situations for humans. That's why principles of both, hardware and cognitive ergonomics, should be applied. One basic principle is controllability (Troy, 1981) which means that the working environment must be transparent and predictable and that the worker must be able to influence it. Models of the working environment and modelling procedures can serve exactly for this purpose. Researchers in cognitive ergonomics are asked to improve the situation by looking for suitable models, modelling tools and modelling procedures. But to date their notions of model, relations between partial models and modelled aspects as well as the purposes of models have not been discussed in a systematic way. A suitable model of the system relevant for HCC research may help to evaluate different proposals, to improve scientific communication and to give hints for further research activities. A uniform representation technique for different but related models might improve communication between researchers, designers and users. Before going into the details of any specific model or representation technique the following questions shall be answered (section 1): What do we mean by 'model'? Which aspects of 'the relevant system' are covered by models? What is the role of 'modelling tools'? In section 2 we show how the focus of interest in HCC has been extended in recent years and illustrate this by means of nets. Today three major roles (as abstractions of agents with similar aims) influence human-computer interfaces: designer, dialog system and user. For the researcher in HCC they determine 'the relevant system', which can be represented by a meta-model. This model is taken as a framework according to which we discuss and classify some models from the literature (section 3). In view of the problems users have with systems and models predefined by others we ask: "When, how, by whom and for whom are models developed and transferred?" The meta-model shows new ways of improving the modelling process. This will be discussed in section 4. Section 5 indicates some open problems and gives directions for further research. Information Systems Design Methodologies And Their Compliance With Cognitive Ergonomy Roland Traunmüller Institut fur Informatik, Johannes Kepler Universität, Linz As methods and tools for the design of information systems have mushroomed in recent years, IFIP has carried out a review process on the methodological development (CRIS Comparative Review on Information Systems Design Methodologies /1,2,3/). Based on the author's participation in the review process a précis of pivotal methodological approaches is given. According to the scope of this symposium four topics relevant to cognitive ergonomics are stressed. So the following sections focus on four subjects’ reasons for the development of methods and methodologies (2,3); pivotal approaches and methods for the design (4,5); topics of product models and process models with relevance to cognitive ergonomics (6,7); information modelling as an example for methodological issues (8). Introducing Statistical Computing Evolution Of The Cognitive System Of The Novice User Gerrit van der Veer Vrije Universiteit, Amsterdam Bert van Muylwijk, Jan van de Wolde T.H. Twente, Enschede, Netherlands The project that is reported in this paper is a pilot study. the main goal is the development of a method to investigate real life situations in which novices receive an introduction to a computer system, determining the interrelations between characteristics of the student; strategy and content of the course; development of a model of the system in the student's mind. If this investigation takes place in an existing course the investigator is normally not allowed to interfere with the teaching method or with the curriculum. The investigation therefore will not have the character of an experiment, the testing of hypotheses will not be a central goal. A description the phenomena is the main result, the teaching strategy ( and the subject matter of the course, the task domain for which the system is introduced) not being a variable in the individual study. The results of these sources may only be speculated upon, albeit that comparison with similar studies in the long run will result in clarity about the relations (Eason, 1983). The authors plan to repeat this study with a number of different institutions for other task domains.