HCI Bibliography Home | HCI Conferences | INT Archive | Detailed Records | RefWorks | EndNote | Hide Abstracts
INT Tables of Contents: 84879095979901030507-107-209-1

Proceedings of IFIP INTERACT'87: Human-Computer Interaction 1987-09-01

Fullname:Proceedings of IFIP INTERACT'87: Human-Computer Interaction
Editors:Hans-Jorg Bullinger; Brian Shackel
Location:Stuttgart, W. Germany
Dates:1987-Sep-01 to 1987-Sep-04
Standard No:ISBN 0-444-70304-7; hcibib: INT87
  1. Invited Lectures
  2. 1. Human Factors in System Development: 1.1 Usability Issues
  3. 1. Human Factors in System Development: 1.2 Psychological Issues
  4. 1. Human Factors in System Development: 1.3 Cognitive Factors
  5. 1. Human Factors in System Development: 1.4 Professionals Workplace
  6. 1. Human Factors in System Development: 1.5 Interface Complexity
  7. 1. Human Factors in System Development: 1.6 Specialized Editors
  8. 1. Human Factors in System Development: 1.7 User Characteristics
  9. 1. Human Factors in System Development: 1.8 Intelligent User Support
  10. 1. Human Factors in System Development: 1.9 Knowledge Representation
  11. 1. Human Factors in System Development: 1.10 User Models
  12. 1. Human Factors in System Development: 1.11 Cognitive Modelling
  13. 1. Human Factors in System Development: 1.12 Programming Tools and Environments I
  14. 1. Human Factors in System Development: 1.13 Programming Tools and Environments II
  15. 2. Design and Evaluation Methods: 2.1 Design and Evaluation Methods
  16. 2. Design and Evaluation Methods: 2.2 Performance Assessment
  17. 2. Design and Evaluation Methods: 2.3 Goals and Guidelines for Design
  18. 2. Design and Evaluation Methods: 2.4 Interface Specification Techniques
  19. 2. Design and Evaluation Methods: 2.5 Dialogue Design and Evaluation
  20. 2. Design and Evaluation Methods: 2.6 Participative Design
  21. 3. Human-Computer Interface Design: 3.1 Human-Computer Interface Design
  22. 3. Human-Computer Interface Design: 3.2 Interface Design of Application Programmes
  23. 3. Human-Computer Interface Design: 3.3 Natural Language Dialogues
  24. 3. Human-Computer Interface Design: 3.4 Evaluation of Input Devices
  25. 3. Human-Computer Interface Design: 3.5 Pictorial Information Presentation
  26. 3. Human-Computer Interface Design: 3.6 Graphics in Human-Computer Interaction
  27. 3. Human-Computer Interface Design: 3.7 Learning and Training
  28. 3. Human-Computer Interface Design: 3.8 Graphical Workstations
  29. 3. Human-Computer Interface Design: 3.9 User Differentials
  30. 3. Human-Computer Interface Design: 3.10 Techniques of Dialogue Design
  31. 3. Human-Computer Interface Design: 3.11 Display Systems
  32. 4. Impact of Computers on Human Behavior: 4.1 Impact of Computers on Human Behavior
  33. 4. Impact of Computers on Human Behavior: 4.2 Organizational Issues of Computer Use
  34. 4. Impact of Computers on Human Behavior: 4.3 Novice Training and Learning
  35. 4. Impact of Computers on Human Behavior: 4.4 User Needs
  36. 4. Impact of Computers on Human Behavior: 4.5 Cognitive Aspects of Information Retrieval
  37. 4. Impact of Computers on Human Behavior: 4.6 User's Language
  38. 5. Forefront Systems and Techniques: 5.1 From the User's Point of View
  39. 5. Forefront Systems and Techniques: 5.2 From the Designer's Point of View
  40. 5. Forefront Systems and Techniques: 5.3 Advances in Design Techniques
  41. 5. Forefront Systems and Techniques: 5.4 Advances in Rapid Prototyping
  42. 5. Forefront Systems and Techniques: 5.5 Advances in Knowledge Based Systems
  43. 5. Forefront Systems and Techniques: 5.6 Novel Application Systems

Invited Lectures

Cognitive Engineering BIBA xxv-xxx
  Jens Rasmussen
Different approaches to the study of cognitive systems can be identified. The AI related 'cognitive science' is based on the information processing metaphor of human cognition in an attempt to reach 'computational' models for behaviour in well-formed micro worlds. Within the field of 'human-computer interaction' studies have been focused on analysis of the communication across the interface between computers and their users. Both these approaches have, quite naturally, been guided by the architecture of present computers. Application of advanced information technology in large scale systems, however, also calls for a more system oriented approach. The paper briefly characterises such a 'cognitive engineering' approach and discusses an approach to analysis and modelling of large scale systems.
On Human Parsing BIBA xxxi-xxxiv
  Friedhart Klix
A model of sentence processing is outlined. It reflects to some extend human strategies in text comprehension. The main differences in comparison with AI approaches are (1) in the immediate access of words (or word groups) to their conceptual representation in memory and (2) in how conceptually organized background knowledge interacts with the text inputs.
How to Design Usable Systems BIBA xxxv-xli
  John D. Gould
Designing useful, usable, likeable computer systems is hard. This is a brief summary of a manual that I am writing to help systems designers and human factors people with this process.

1. Human Factors in System Development: 1.1 Usability Issues

Objectives for the Humanization of Software - A New and Extensive Approach BIBA 5-10
  Helmut Balzert
If software is to be designed ergonomically, the first step must be to clearly define all objectives. These can be derived from the science of ergonomics and from cognitive psychology. A total of 23 objectives have been listed, which can be divided into four groups: development of personality, acceptability, support of human-computer-interaction, and support of human information processing. The relations of the objectives to one another is shown in a dependence diagram. As an example for operational transformation, one objective has been divided in a number of dimensions. Three levels of elaboration have been specified for each dimension. A method for defining objectives for a concrete software development can be derived from the General Objectives Model. Objectives should not only be determined for software construction, but also for the purpose of evaluation and standardization.
The Presentation of Human Factors to Designers of I.T. Products BIBA 11-16
  Margaret Galer; A. J. Russell
In order effectively to provide human factors to the designers of IT products the presentation and content of the human factors must suit the requirements of the user, for the task that he/she wishes to carry out, in their normal working environment. The ESPRIT-HUFIT project is developing methodologies for integrating human factors into the IT product design process. ICL and HUSAT are concerned with the specification and prototyping of the tools to deliver the human factors methodologies. The development of a knowledge based Decision Support System (INTUIT) is described as a working example.
The Dialectic of Usability Engineering BIB 17-20
  John Whiteside; Dennis Wixon
Developing a User Interface Technology for Use in Industry BIBA 21-26
  John L. Bennett; Douglas J. Lorch; David E. Kieras; Peter G. Polson
We are developing a user interface technology to address ease of learning and ease of use concerns on the user side of the interface during the design process. Modelling the user how-to-do-it knowledge required by a design is one step toward development of such a technology. We report on an evolving methodology that is intended to give developers early warning indications of potential usability problems that may arise from a set of design decisions.

1. Human Factors in System Development: 1.2 Psychological Issues

Mastering the Complexity of Dialogue Systems by the aid of Work Contexts BIBA 29-33
  Wolfgang Dzida; Claus Hoffmann; Wilhelm Valder
For user interfaces of complex application systems we develop a component securing a sufficient transfer of knowledge about the application of the complex system. The knowledge of experienced users is to be analyzed and provided such that less experienced users are enabled to use the rich functional spectrum. Otherwise, the full utilization of the functional spectrum and the efficiency of system use remain to be jeopardized. The concept presented here can be applied to complex constructional workstations for CAD and in software development if these workstations have been developed on UNIX basis.
Do Users Know They Have User Models? Some Experiences in the Practice of User Modelling BIBA 35-41
  A. G. Sutcliffe; A. C. Old
This paper describes practical experiences of user modelling UNIX to specify a direct manipulation interface. Data capture employed questionnaire, system logs and interviews. Results are discussed in light of the utility of various user modelling concepts.
Does Computer Interest Induce Mechanical Thinking? BIBA 43-47
  Yvonne Wærn
Many people are concerned about the effects of the computerization. Here one of the worries will be attended to dealing with the effect of computers on people who work with computers: are the computer experts getting like machines by their intercourse with computers? Two studies are presented, aiming at analyzing eventual differences in thinking between people who are interested in computers and people who are not familiar with computers. In the first study, mathematical and logical tasks were used, where a positive transfer effect from computer interest was expected. In the second study, creative and verbal tasks were added, in order to study the eventual negative effect of computer interest. Students in computer science were compared to students in psychology. In the first study it was found that computer science students used logic and formal reasoning to greater extent than psychology students did. In the second study psychology students excelled in practical creativity.

1. Human Factors in System Development: 1.3 Cognitive Factors

The Use of Thinking-Out-Loud and Protocol Analysis in Development of a Process Model of Interactive Database Searching BIBA 51-56
  Thomas T. Hewett; Sari Scott
A variation of the "thinking-out-loud" and protocol analysis method of studying problem solving behavior was used to observe the interactive searching of experienced bibliographic database searchers. To control for the effects of different search problems, three different search requests were structured to represent major generic kinds of search problems. Analysis of pre-search protocols, on-line search transcripts, and post-search debriefings of nine searchers, three of whom were randomly assigned to each type of search request, revealed that, while there were differences in searcher behavior as a function of type of search problem, there were important similarities in searcher behavior. These similarities made it possible to develop a general characterization of the process of database searching. This process model was subsequently used to structure tutorial exercises used in training beginning searchers. These tutorial exercises were embedded in an on-line search assistance program. Evaluation studies conducted both on the training exercises and on the search assistance system as a whole supported the idea that the search process model developed through "thinking-out-loud" and protocol analysis was a reasonable representation of the search process and was useful in training searchers.
Models in Human Computer Interaction: A Classification with Special Reference to Their Uses in Design BIBA 57-63
  Andy Whitefield
Models are common in the human computer interaction (HCI) literature. This paper proposes a classification of HCI models according to who is doing the modelling and who or what is being modelled. This allows similarities and differences between the models to be identified. The classification is then used to consider one of the functions of HCI models - the ways in which they might be used in computer system design.
Parsing and gnisrap: A Model of Device Use BIBA 65-70
  T. R. G. Green; R. K. E. Bellamy; J. M. Parker
It is obvious that interactive environments make some tasks easier to achieve than others, but less obvious why. This paper introduces a model of coding that highlights features of the device, task, interaction medium and user knowledge that are important in determining the ease of use of a support environment. The model has been implemented in Prolog and applied to the domain of expert programming. Although the model is still at an early stage of development, it clearly shows the need to build device languages and support environments which complement each other, in the light of users' tasks and knowledge structures. Ways of achieving this aim are discussed.
An Empirical Validation of Cognitive Complexity Theory BIB 71-75
  P. H. Vossen; S. Sitter; J. E. Ziegler

1. Human Factors in System Development: 1.4 Professionals Workplace

The Human Interface to Computerised Banking Services BIBA 79-84
  Dermot Scallon
The use of computers to provide automated financial services is poised to become one of the most exciting major growth areas throughout the developed world. Computer-based financial services involve such technology as Automated Teller Machines, (ATMs) video-disk systems, "Smart Card" and "Laser Card" technology, screen-based and voice-based home banking, and Electronic Funds Transfer at Point Of Sales (EFTPOS) systems. While current generation ATMs have already introduced many people to computer-based services, the above technologies will require increasingly sophisticated understanding of computer systems to ensure firstly, that they are used at all, and secondly, to ensure the maximum beneficial returns for both customers and vendors. These technologies will play a major part in introducing non-computer-users to a role in which they will increasingly find themselves, that is how to utilise the computer to obtain a desired service.
Statistical Software and the User Interface BIBA 85-88
  Michael E. Dewey; Ann G. Harding
This paper reviews our experience using statistical software and counselling doctors and psychologists in its use. We argue that the interface to such software has remained uninfluenced by modern methods, and this has contributed to the difficulties which users find with such programs. We illustrate these difficulties with a series of examples from the range of machines and packages which we have used. We present the taxonomy which we have developed to classify these difficulties, which are categorised under two main heads: the myth of the user and the mentality of the batch era. The paper concludes with a description of the interface which we have developed based on this taxonomy, and an outline of the suite of statistical programs written in Pascal for small microcomputers.
Integration at a Work Place for Statistical Consulting BIBA 89-92
  G. Dirlich; H. Federkiel; E. Hansert; A. Yassouridis
We are developing an integrated work place for the statistical consulting of research projects. Typically, in a consulting situation a scientist who has encountered methodological, and in particular statistical, problems in his research meets with an experienced statistician and discusses the problems. The work place shall provide support for this consulting dialog.
   In the first part of the present report, we outline a problem treated in a consulting dialog and a task structure developed by the client (scientist) and the consultant (statistician).
   In the second part we discuss a typical tool at the work place with respect to its function, its interaction with other tools, its surface and mode of operation.
   The third part is a brief presentation of some design guidelines which have emerged in previous prototyping cycles.
Data-Base Organization and Cognitive Structure: Using Information Systems Organized by Oneself and by Others BIBA 93-98
  Frank Henry Piekara; Gerhard Strube
New data are reported on the relative efficiency of database descriptor systems devised by either the user or by others. Former results suggested that performance in information retrieval was almost twice as high for systems organized by oneself in comparison to systems organized by others. In contrast, we find that (1) individual differences amount to less than one third of the elements of domain specific cognitive structures, and (2) the disadvantage of using descriptor systems organized by others is rather small (about 25 percent) and statistically insignificant.

1. Human Factors in System Development: 1.5 Interface Complexity

Designing a Human Interface by Minimising Cognitive Complexity BIBA 101-108
  Ian A. Clark
A schematic notation (the "Function/Object/Relationship/State" or FORS Diagram) is developed on an elementary example of a human computer interface. Published research findings into the causes of novice users' errors and difficulties with interactive systems are reinterpreted in terms of this view, leading to the representation of cognitive complexity in terms of topological features of the diagram. A design methodology is proposed whereby, starting with a simple FORS diagram, functionality is added and the diagram is re-drawn to reduce cognitive complexity. The technique is demonstrated by a case-study to design a "listor", a type of powerful interactive utility. The resulting "listor" is compared with examples of listors in actual marketed systems.
A Quantitative Measure for the Complexity of Man-Machine Interaction Process BIBA 109-116
  Klaus Kornwachs
The reliability of man-machine interaction process depends in a strong manner from its complexity. Since all complex systems can only be described as hierarchical ordered subsystems or subprocesses, the man-machine interaction is described as such a complex system. The difference between an algorithmic and a cognitive subprocess is pointed out. A quantitative measure for the complexity is developed and its consequences are applied to Cognitive Complexity Theory. The proposed measure is able to indicate conditions under which the reliability of learning, transfer and performance can not predicted anymore.

1. Human Factors in System Development: 1.6 Specialized Editors

A Syntax-Directed Graphics Editor BIBA 119-124
  Eric Inman
When editing information in graphics form, where the graphics are governed by a graphics syntax, the user should not be required to draw the graphics representation, but only to specify the information that is to be represented. The editor should then use the syntax rules to convert that information to its graphics form automatically. This not only ensures that syntactically correct graphics are produced, but also frees the user from the time-consuming task of drawing. To put it another way, a graphics editor which incorporates a knowledge of the syntax of the graphs to be drawn should provide many of the advantages that syntax-directed structure editors provide to software developers. These considerations guided the design of CGEDIT, a syntax-directed editor for composition graphs.
STRUEDI: A Lisp-Structure Editor for Novice Programmers BIBA 125-129
  Axel Kohne; Gerhard Weber
Due to the uncommon syntax of LISP, novice programmers often get into problems. These problems do not only result in syntactical errors but also in many semantic, algorithmic, and even planning errors. Working memory overload is considered to be one major reason for these high level errors. STRUEDI, a structured, syntax-directed LISP-editor is designed to reduce these problems. The user is guided to program in a top-down-style by successively putting LISP-constructs into slots of superior constructs. Constructs can be typed in from a menu which contains already learned or defined LISP-procedures or can be typed in from the keyboard. Using predefined constructs only syntactically correct programs can be built with STRUEDI: Errors are flagged and explained. Additionally, programmers can get help on demand. A first evaluation study indicated that working with STRUEDI increases the performance in applying the knowledge about syntax and semantics of LISP.
User Modeling for Syntax-Directed Editors BIBA 131-134
  Lisa Rubin Neal
Design principles for user interface and system design are clearly needed, but no sound design principles will emerge until user cognitive styles and skills are characterized by adequate user models. We selected syntax-directed editors as a testbed for our research, primarily because we were interested in the reasons underlying their lack of success and in ways to increase their effectiveness. We formulated a five-dimensional user model, of which three dimensions are related to expertise and two dimensions are related to learning and decision-making styles. The user model shows relevancy to syntax-directed editor and other interface designs, and we found evidence of non-trivial interaction between the user model and system design.
GEGS - A System for Generating Graphical Editors BIBA 135-141
  Gerd Szwillus
Graphical hardware is getting cheaper, faster and more powerful, whereas the process of using the available facilities in several fields is rather underdeveloped. The programming of graphical user interfaces is a very laborious job. Compared to the design of good "textual" interfaces, a much larger number of details have to be taken into account. We present a generator-based approach for creating graphical system attacking this problem. This paper gives an insight into the basic concepts of our approach. Two aspects are considered particularly interesting in this context: First, the characterization of the class of pictures, editors can be generated for; and second, the design of the generator input language.

1. Human Factors in System Development: 1.7 User Characteristics

Individual Differences in Human-Computer-Interaction: How Can We Measure if the Dialog Grammar Fits the User's Needs? BIBA 145-149
  Thomas Greutmann; David Ackermann
The human-computer interaction for two calculator tasks and one drawing task was investigated. It was found that individual differences exist and have to be considered in system design. Methods that indicate individual preferences and difficulties with task structuring and single commands are presented. Results of one task domain turned out not to be transferable to other domains. This questions the generality of results in human-computer interaction.
User Knowledge Evaluation: An Experiment with UNIX BIBA 151-156
  Michel C. Desmarais; Michael Pavel
The building of interfaces that adapt to the style and expertise of a given user is currently a major challenge in human-computer interaction. Two of the main difficulties have to do with (1) how to represent users' knowledge of a subject content and (2) how to evaluate a given user's level of expertise unobtrusively? In this paper we describe a tool for building adaptive interfaces on UNIX that addresses these two difficulties. The first component of this tool assesses a given user's expertise from the observation of his/her interaction with the system. It uses an "overlay model" for the representation of user expertise, that is, its representation of a user's expertise is a subset of a global set of knowledge items. The set of knowledge items is structured according to interdependencies among the items. This enables inferences to be made during the knowledge assessment process. The second component of the tool builds such a knowledge structure from data on a number of user's expertise.
A Psychological View of "User-Friendliness" BIBA 157-163
  Andrew Dillon
This paper analyses human-computer interaction within a cognitive psychological framework. On the basis of an in-depth survey and longitudinal experimental study, the manner in which the style of an interface influences user performance and knowledge acquisition is discussed. The model proposed in this paper suggests that the user can meaningfully be viewed as an information processor who actively extracts information from the interface. The quality of information available for extraction is seen as having a formative effect on conceptual model development and knowledge gain. From this perspective the nature of "user-friendliness" or usability is seen as multiply determined according to the information needs of the user. Thus users are distinguished according to psychological criteria of progressive knowledge/skill development. Empirical evidence supporting this view is presented.
Cognitive and Social Models of the User BIBA 165-169
  G. Nigel Gilbert
A user model is a description of the user created or selected by a program to enable it to adapt its behaviour to the user. From a brief review of current attempts at user modelling, three broad types of model are distinguished: those based solely on the user's owe explicit self-descriptions ('direct' models); those which try to fit the user into one of a number of stored stereotypes ('stereotype' models); and those which attempt to model the user's intentions and plan ('intentional' models). The adequacy of these approaches to user modelling is evaluated according to the degree to which they are capable of faithfully representing relevant aspects of the user. It is suggested that even the most sophisticated of the approaches, the intentional, falls a long way short of what might be achieved. This is because the intentional approach is based on modelling the assumed cognitive state of users, including their plans, intentions and goals. Drawing on studies of human-human interaction, it is argued that the cognitive states of other interactants are not available to a speaker although adaptation to others' conversational moves appears to be successfully achieved. It is therefore suggested that current attempts to construct better models of the user's cognitive state are misconceived, and that attention should focus instead on providing programs with models of the system itself and models of the interaction, to allow reflexive reasoning and meta-level commentary on the user system dialogue.

1. Human Factors in System Development: 1.8 Intelligent User Support

Information and Consultation Systems - A New Dimension of User Support BIBA 173-178
  Helmut Balzert; Rainer Lutze
Providing information, consultation and recommendation services to the user of a software system is not only a subsidiary task of a help component within a software system, but also the original task of many software systems, for example: databases or management information systems (MIS). How far is an application independent realization of such different and cost intensive services possible by a common concept? We present a novel architecture of software systems capable to handle both tasks. The rationale behind the architecture is to have a common man-computer interface (MCI) and a common information, consultation and recommendation system (ICRS) tailored to communicate with different application systems (AS). By applying this architecture, a substantial cut-back in the development costs of a software system could be achieved.
Operating Systems Support for Flexible Human-Computer Interaction BIBA 179-183
  P. Brossler; K. Faszl; B. Freisleben
In this paper it is argued that suitable operating systems support can be beneficial for modelling the interaction between human users and the computing system. The proposed solution is based on a consequent use of the object model and the availability of lightweight processes. The flexibility of our approach is demonstrated by several examples.
The Diagnosis of User Strategies BIBA 185-189
  Michel C. Desmarais; Serge Larochelle; Luc Giroux
This paper presents the architecture of an expert system for the diagnosis of user's strategies on a text-editing task. The main objective of the system is to recognize inefficient strategies and to suggest more efficient ones. We demonstrate how strategies can be represented with a formalism called procedural network, and how they can be identified by a standard parsing procedure. The result of the parsing process is a hierarchy of goals, sub-goals, and observable actions at the terminal nodes, which serves as input to a knowledge assessment module. Which actions or sub-goals are chosen to accomplish a goal enables inferences to be made by the knowledge assessment module on what is presumably known or not by a particular user. Suggestions of what skills should be improved can be made on the basis of this knowledge assessment.
Do People Really Use On-Line Assistance? BIBA 191-194
  Thomas Moll; Roland Sauter
In an interdisciplinary research project, psychologists and computer scientists investigated the real user behavior of 24 tool and die makers using an interactive CAM system. Data were collected in a field study on the handling of a context-specific help system by means of a combination of techniques such as on-line questions, logfile recording, thinking aloud, video self-confrontation. The advantages and disadvantages of such on-line assistance will be described here.

1. Human Factors in System Development: 1.9 Knowledge Representation

A Framework of Developing Semantic Models of User Performance BIBA 197-202
  Herbert Hagendorf
The development of models of user performance in man-computer interaction must take increased account of the representation and organization of knowledge. This paper describes a conceptual framework of this kind of the basis of psychologically based model of knowledge representation. Attention is centered on operative knowledge units that are learning-dependent and exercise an influence on both the planning processes and the specification of action sequence for the computer. The results of investigations made on the basis of a simple graphic system will be interpreted in this very framework.
Complexity Problem Spaces: Modelling the Knowledge Needed to Use Interactive Devices BIBA 203-208
  Stephen J. Payne
A model is presented of the way users construct device-oriented problem spaces. The core of the model is a complex, layered problem space, consisting of two state spaces, the goal space and the device space, and a semantic mapping between them. Structural redundancy is identified as a formal property of such problem spaces that predicts the learnability of devices. Within the framework, two models of learning are described which lead to qualitatively different versions of the problem space: Operational accounts treat primitive interactions with the device as syntactic components of useful methods; Figurative accounts elaborate the conceptual model of the device to provide semantic interpretations of the primitive interactions.
The Analysis of Knowledge Representation of Nuclear Power Plant Control Room Operators BIBA 209-214
  Pierre Alengry
In this paper we present an experiment carried out on nuclear power plant operators. The aim of the experiment is to represent the knowledge the operators use in their reasoning. We offer an initial formalization of the knowledge of the system state transformation rules in the dorm of Causal Relations Chains (CRCs). CRCs have an explanatory function in action and goal setting. Our hypothesis is that the operators build up mental models of the system functioning, made up with CRCs. We then show that the CRCs evoked by the operators are based on underlying knowledge structures.
Representation of Domain Knowledge in an Intelligent Help System BIBA 215-220
  Christel Kemke
The SINIX Consultant (SC) is an intelligent help system for the SINIX operating system. It is supposed to answer natural language questions about SINIX concepts and commands and also give unsolicited advice to the user. The basis, in order to fulfill these tasks, is a rich knowledge base with respect to the SINIX system which reflects the technical aspects of the domain as well as a user's view and her use of the system.
   The SINIX Knowledge Base consists of a taxonomical hierarchy of concepts according to different views or classifications of the domain concepts. Domain concepts are commands and (virtual) objects of the SINIX system; higher level concepts correspond to natural language terms, mental model entities, or more general abstract actions and objects. A single concept is described with respect to its function, structure, use and/or relation to other concepts. The main emphasis is on the representation of commands, which includes characterizations of their application and function used for user-adequate explanations. The main ideas of a formal description of their semantics, necessary in order to enable reasoning and problem solving processes, are outlined.

1. Human Factors in System Development: 1.10 User Models

The Space-Concept and the Control of Space BIBA 223-227
  Hans Kohler
The users of computer screens do often not know the underlying principles for the designers concept for model space or screen space. This paper gives some names to important things within this scope Some metaphors are discussed: rectangular sheet, sawtooth sheet, continuous band, open sheet, projected spiral and miscellaneous surface. The possible actions on the screen are crucial sources for the understanding of the screen space concept. The curser is the transitional object for that actions. Six recommendations to the developers of screen oriented systems close the paper.
Embedded User Models BIBA 229-235
  Dianne M. Murray
A global definition of user models is presented and the characteristics which define a particular form of model (an 'embedded user model') are highlighted and discussed. The distinctions between various forms of conceptual, application and student models are described and examined. These definitions are located within a general categorisation of user modelling techniques, based on implementation methodology, and on 'ownership' of the model itself. An extensive literature review of recent research in modelling techniques is referenced and a number of techniques described. The architecture of embedded user models is contrasted to that of User Interface Management Systems and an outline design specification for a system encompassing such a model is proposed.
The Role of Task Characterisation in Transferring Models of Users: The Example of Engineering Design BIBA 237-243
  Clive Warren; Andy Whitefield
This work concerns the modelling of engineering designers, to be used as an input to Computer Aided Design (CAD) system development. The paper describes ways of characterising design tasks, and reports the transfer of an existing model to a new task. The characterisation serves the transfer by allowing selection and comparison of tasks, and generalisation to new tasks.
System Adaptivity and the Modelling of Stereotypes BIBA 245-253
  David Benyon; Peter Innocent; Dianne Murray
The argument for adaptivity in a system is developed and related to previous theoretical work on adaptive interface design. We attempt to integrate recent research findings with practical experimental trials to provide a new formalism for system adaptivity. The experimental vehicle is a small CBT/tutoring system which incorporates embedded models of individual characteristics and student information in the form of 'stereotypic' attributes and user profiles. We describe the system characteristics and operation, give our experimental results and detail future planned work. We discuss some of the implementation difficulties already encountered and those which we expect to be apparent in future systems.

1. Human Factors in System Development: 1.11 Cognitive Modelling

Analysis of Cognitive Activities in Process Control for the Design of Computer Aids. An Example: The Control of a Blast Furnace BIBA 257-262
  Jean-Michel Hoc
This paper presents the preliminary findings of a research program designed to model diagnosis activity in order to propose a computerized support system for a continuous process characterized by long response latencies: the blast furnace. Part one examines process control from a cognitive point of view. This is followed by methodological considerations. The major features of the control activity are then defined in terms of its main components: supervision, diagnosis, and intervention. Conclusions as regards computerized support system design are put forward.
CATOOL: A Computer-Based Tool for Investigations of Categorical Information in Mental Models BIBA 263-267
  Svend Erik Olsen
Automated probing of mental models is an area of growing importance. Currently, we are developing a computer-based tool which will be able to elicit information about the user's category representations in a given domain. The tool will incorporate a number of techniques for research on categories and categorizations, and it will preserve the dynamic and context-dependent character of the user's category representations. In addition to elicitating categorical information, the tool can also be used in instructional contexts, where it can provide the user with metacognitive feedback.
Mental Model and Procedural Elements Approaches as Guidelines for Designing Word Processing Instructions BIBA 269-274
  Franz Schmalhofer
By investigating novices' learning and performance, the effectiveness of instructions which describe the structure of a computer system and how it works was compared to the effectiveness of explicit how-to-do-it instructions. The experimental results suggest that mental model guidelines should be used for designing instruction materials.
Analysis-Based Learning in Human-Computer Interaction BIBA 275-280
  Clayton Lewis; Stephen Casner; Victor Schoenberg; Mitchell Blake
A model based on recent advances in machine learning can shed light on how people learn about unfamiliar systems from demonstration. The model uses simple heuristics to assign causal roles to user actions in a demonstration, and then forms new procedures for related goals using this analysis. Empirical studies have provided support for the general framework of the model, though many important specifics are unresolved. The model and supporting results provide some guidance for the design of systems that will be easy to learn from demonstrations.

1. Human Factors in System Development: 1.12 Programming Tools and Environments I

A (Formal) Model for (Iconic) Programming Environments BIBA 283-290
  Ephraim P. Glinert; Jakob Gonczarowski
Our objective in this paper is to motivate and develop a formal model for iconic programming environments. Our model for these highly visual environments is based on the concept of class-instance pairs, and is therefore named CLIP. After the exposition, we enumerate several important applications which attest to the model's utility and broad scope.
APT: A Principled Design of an Animated View of Program Execution for Novice Programmers BIBA 291-296
  Tim Rajan
The research described here is concerned with the principled design of a computational environment which depicts an animated view of program execution for novice programmers. These design principles are aimed at solving the problems that novices face when learning new programming languages, and have been embodied in an Animated Program Tracer (APT) for Prolog. The aim of this research is to develop a more systematic, if not yet scientific, basis for the design of animated tracing tools.
Discourse Rules in Program Comprehension: Emergence of a Construct Affordances Rule? BIBA 297-302
  Laura Marie Leventhal
Several investigators have suggested that rules of discourse for programming exist. These rules reflect programmer expectation and practice. Violations of these rules of discourse may have a negative impact on program comprehension, particularly among expert programmers.
   One general type of discourse rule is a Construct Affordances Rule. This rule states that a particular programming construct is appropriate for and should be used in certain situations. Soloway and Ehrlich [1] demonstrated that one instance of the Construct Affordances Rule had an effect on performance in a programming comprehension task, particularly among expert programmers.
   In the current study, both comprehension measures and interest measures were collected from novices, intermediates, and experts. The effects of instances of the Construct Affordances Rule involving the WHILE, FOR, IF, and REPEAT Pascal constructs on comprehension were measured using a fill-in-the-blank procedure. The results revealed that the discourse rules may influence programmer behavior but not necessarily in the predicted ways. However, expertise level had the predicted effect on comprehension. These results are discussed in terms of the role of discourse rules in the formation of problem representations, especially among experts.

1. Human Factors in System Development: 1.13 Programming Tools and Environments II

The UNIVERSE Program Development Environment BIBA 305-309
  Jeff Parker; Bob Hendley
This paper reports on research into the investigation and development of facilities for user oriented program development aids. The paper describes a fully operational programming environment, UNIVERSE, which incorporates recent ideas from two areas of research. In particular UNIVERSE provides more ergonomic tools for the programmer (such as structured editors, incremental compilers, animated debuggers), and it uses recent work on the psychology of programming, in particular programming plans (Soloway [1], Johnson [2], Spohrer et al [3], Rist [4]), and ideal notational design, (Green [5], Gilmore [6]).
   The role of a program development environment can be divided (rather abruptly) into increasing the functionality of the workstation, by supplying automated facilities for program construction, compilation, module control, execution, etc., and improving the usability of the system by resolving psychological difficulties in the use of those aspects of UNIVERSE that are particularly concerned with programming psychology. Research findings on typical programming tasks, such as program comprehension, have repeatedly demonstrated the shortcomings of existing notations considered purely as textual communication devices, a major design aim of UNIVERSE is to create an environment in which these shortcomings are overcome.
   Environments such as UNIVERSE also present the opportunity for the development of interactive programming languages which are no longer subject to the constraints imposed by traditional software development tools.
The User Interface of PSG Programming Environments BIBA 311-315
  Rolf Bahlke; Manfred Hunkel
The PSG system is a generator for interactive language-specific programming environments. A generated environment consists of a language-based hybrid editor, a library system, and an interpreter augmented by a debugging system. The user of a PSG generated environment communicates via a uniform interface with any of the above system components. This paper presents in detail the user interface and its design considerations of the PSG implementation running on ICL-PERQ and pcs-CADMUS workstations equipped with a raster-graphics display and a pointing device. The environment and its user interface have been designed to support both casual and experienced users. In order to achieve this goal, the environment is devoid of any command language: to trigger a certain task, the user simply has to select the appropriate item from a menu of alternatives. The editor allows structure-oriented editing as well as conventional text editing. Although the user interface has been primarily designed to be driven by PSG environments, it could nevertheless be employed by other text-processing software. In addition to a screen-oriented text-editor, the interface provides windowing, static menus and pop-up menus, and access to the trigger buttons of the pointing device.
Graphics and Learning: A Study of Learner Characteristics and Comprehension of Programming Languages BIBA 317-322
  Nancy Cunniff; Robert P. Taylor
This paper reports empirical research supporting the intuition that visual representation can make at least some abstract concepts easier to comprehend. A study of novices' comprehension of computer programs was conducted using two programming languages, FPL (a graphically represented language) and Pascal (a textually represented language). The study focused on whether some identifiable subset of learners performs more a quickly and/or more accurately on comprehension tasks when reading program segments coded in the graphically represented language than when reading their equivalent in a textual language. The findings support the intuition that graphical representation results in more rapid and accurate comprehension than does textual representation. This bias is even more marked for the visually apt learner. The findings are of importance to educators at all levels, suggesting that, in the design of learning materials, the graphic capabilities of computing ought to be more fully exploited to meet the differing needs and learning profiles of students.

2. Design and Evaluation Methods: 2.1 Design and Evaluation Methods

Multi-Service-Terminals -- Human Factors Studies with an Experimental Prototype BIBA 327-332
  Angela Prussog; Werner Blohm
Service-integrated networks suggest a concept of accessing various services from one single terminal unit ("multi-service-terminal"). Due to the variety of functions, several problems have to be expected for the user in the course of operating such a terminal. In order to provide empirically based recommendations for a 'user-friendly' design of multi-service-terminals, several design concepts were developed. The significance of these concepts was evaluated within human factors investigations with an experimental prototype system, which gives access to various narrow- and broadband telecommunication services.
   Within a pilot study two different types of keypads -- with hard and softkeys -- were tested and evaluated by non-expert subjects. Results of the pilot study show that even inexperienced users were able to manage the operating procedures successfully, if multi-service-terminals are designed adequately. Apart from the advantages of keypads with softkeys, especially with respect to their flexibility, certain problems and disadvantages for the user became evident.
Good Software Design: What does it Mean? BIBA 333-335
  Ernest Edmonds
Much has been written about software design and many principles have been proposed. The paper briefly reviews the current views and draws from them a proposal that might help to define 'good software design'. The proposal considers all users, including maintenance engineers, installing engineers, etc., as well as the end user, and develops the notion of small coherent sets of principles that are used as the basis for each design decision. It is important to notice that these sets need not be the same for each class of user. The influence that these ideas might usefully have upon design is discussed.
Usability Evaluation and Feedback to Designers -- An Experimental Study BIBA 337-340
  F. Novara; N. Bertaggia; N. Allamanno; S. Fox; W. Olphert
A major goal of the ESPRIT HUFIT Project is to identify usability criteria and develop measures of usability which will in themselves be useful to designers of IT systems. This paper reports an experiment conducted as part of this Project which aimed to pilot measures of usability. The aim of this experiment was to assess the usability of a newly developed text processing package. Twenty six subjects were required to practice with the package to a given standard, then were asked to perform an experimental task. Various measures, including time, errors and help requirements were taken, as well as subjective evaluations of the package. The data from this experiment were then analysed to identify a number of areas where there were actual or potential usability problems with the product. In a second phase of the work, the experimental results were elaborated in a simple format and communicated to the product designers.
A User Centred Approach to the Design of a Knowledge Based System BIB 341-346
  K. D. Eason; S. D. P. Harker; P. F. Raven; J. R. Brailsford; A. D. Cross

2. Design and Evaluation Methods: 2.2 Performance Assessment

Methods for the Ergonomical Evaluation of Alphanumeric Computer-Generated Displays BIBA 349-353
  W. Graf; P. Elsinger; H. Krueger
Using ten differently structured display formats containing the same information, the relationship between geometrical characteristics of a display, search time, subjective rating and eye-movement was investigated. The results of Tullis (1984) could be confirmed. There is a high correlation between the geometrical characteristics of a display and the search time for an item on the display as well as the subjective rating. Therefore it is possible to predict search time and subjective rating according to geometrical characteristics. Further it could be shown, that the measured eye-movement parameters i.e. saccade amplitude and fixation time correlate partly with the geometrical parameters. The eye-movements give also a clue about the exact position of the problems in the displayed format.
Cognitive Efficiency During High Work Load in Final System Testing of a Large Computer System BIBA 355-360
  Anita Rissler; Lena Jacobsson
In a longitudinal field study, testing engineers were studied when completing a computer project with a tight deadline. Focus is on cognitive efficiency towards the end of a long period of high work demands. Two tests of complex cognitive processes were repeatedly administered during the final month of computer systems testing and after the following summer vacation. The results indicated that cognitive efficiency could be maintained when it was most needed at work, but analyses of coping strategies showed, that the price for good performance was high, involving psychological, biological and social costs to the employees.
"Generics" in Human Decision Making BIBA 361-365
  Sebastiano Bagnara; Antonio Rizzo; Franca Stablum; Michele Visciola
In the early phases of designing a human/automation system, strategic decisions are taken as those concerning responsibility allocation and decision making. In order to have ergonomic expertise interacting at the proper level when such strategic decisions are negotiated, human compatibility should be used as a criterion for evaluating those decisions. The present paper describes how to analyze the cognitive decision making activity to be assigned to the human operators on the basis of decisional programs and units and how to evaluate such activity as for cognitive complexity and mental workload. Conclusions from a case study will also be reported.
Methodologies Employed in the Psychological Evaluation of H.C.I. BIBA 367-373
  Marian Sweeney; Andrew Dillon
This paper presents a review of a range of techniques employed in a psychological evaluation of human performance and skill development in computer interaction. The format outlines the authors' experiences in applying methodologies appropriate to the nature of the task under investigation and to the experimental design.
   It is all too easy to find after extensive time and financial investment that the data is spurious and difficult to analyse, that in a complex evaluative situation you have lost sight of the issue or at the worst that the research has not even addressed the real issue. The central promise of the paper outlines why and how the research interest must be explicitly defined at the outset in order to design and apply the appropriate methodology before any data capture begins.
   While the studies reported investigate two quite distinct issues in H.C.I., (data base usage and development of programmer skill and knowledge), it was found that many of the techniques reviewed or employed could be as effectively applied to many of the issues of interest to human factors investigators.

2. Design and Evaluation Methods: 2.3 Goals and Guidelines for Design

Giving Control Back to the User BIBA 377-382
  Alan Dix
Interactive systems inevitably reflect the structures of the language in which they are written. Imperative programs gave a pre-determined thread of control and the resulting dialogues are often over-determined, in the sense that they do not allow users to structure the dialogue in ways natural to them. Functional languages potentially have a freer choice of control flow. Techniques are proposed whereby these choices are reflected in the resulting dialogues. This will lead to interfaces where the user exercises more control over the dialogue.
   The aim is to make the user's choices in the dialogue correspond to the choice of evaluation order of the functional program, leading to a structural correspondence between the two. To achieve this various additions and constraints are needed, input-output primitives are treated as non-deterministic functions and the choice of a lazy evaluation strategy means that the only sequencing between these primitives arises from the essential data dependencies. Where additional sequencing is required, it must be added explicitly using special results returned from each primitive called completion events. These are also used to provide real time support for waits, time-outs etc.
   These techniques are demonstrated using various examples from information systems. By making designers explicitly introduce any non-essential sequencing, they are encouraged to produce systems that more faithfully reflect the freedom of sequence offered by such techniques as forms-oriented input-output and multi-windowing, thus giving control back to the user.
Cognitive Processes in Software Design: Activities in Early, Upstream Design BIBA 383-388
  Raymonde Guindon; Herb Krasner; Bill Curtis
The first goal of this paper is to present some of the main activities occurring during early, upstream software design by experienced designers. We concentrate on the variety of strategies found between designers. The second goal is to bring to attention certain cognitive activities in design that have not been observed or emphasized in other studies of design, such as serendipity and the process of understanding and elaborating the requirements through exploration of the designer's mental model of the problem environment. These activities are likely to be critical in upstream design.
Conceptual Consistency in the User Interface: Effects on User Performance BIBA 389-394
  Wendy A. Kellogg
Consistency is often promoted as an important goal of user interface design, yet few definitions exist, particularly in a form suitable for guiding design. Most design guidelines stress procedural consistency, e.g., low-level interaction techniques, or the assignment of programmable function keys. This paper argues that interface consistency must also be addressed at a conceptual level. Conceptual consistency is defined in terms of the internal coherence of a system's structure and the nature of the mapping from user task goals to system procedures. Interfaces to an automated office system were constructed in which low-level interaction techniques were held constant, but which varied the degree of conceptual consistency. The results of a study in which users performed representative tasks on these systems is reported and discussed with respect to the importance of various kinds of consistency in the interface.
Top-Down Interactive Systems Design: Some Lessons Learnt from Using Command Language Grammar BIBA 395-399
  B. D. Sharratt
Approaches to interface design often adopt an analytic method working in a top-down fashion from a task analysis through various levels of detail. Moran has suggested that his Command Language Grammar (CLG) can be used as such a design tool, with the interface specified at four levels and each level providing a complete description of the interface.
   The paper describes a design exercise where postgraduate students used CLG to specify interfaces for a transport timetabling system and then prototyped their designs. Following a detailed analysis of the CLGs produced, areas of difficulty such as the syntactical structure of CLG, the propagation of initial design problems at lower levels of the grammar and the reliance on complex "downward" mapping across the levels are discussed. Extensions to CLG, including automatic checking of consistency and mapping between levels, and the incremental addition or alteration of CLG structures, are outlined. The issues encountered in the CLG specifications are used to demonstrate the problems associated with a top-down design process; and indicate additional features which would provide more adequate support for such a process.

2. Design and Evaluation Methods: 2.4 Interface Specification Techniques

Modular Specification Methods for User Interfaces BIBA 403-408
  T. T. Carey; C. H. Graham
A specification model for user interfaces is illustrated by examples from a computer conferencing system and a personal computer paint package. The examples show formal methods for describing the semantics of application-independent interaction styles, and for condensing a specification using designers' conceptual abstractions.
Formalising Models of Interaction in the Design of a Display Editor BIBA 409-414
  A. J. Dix; M. D. Harrison
Many of the conceptual properties often suggested to be valuable of text editors are either device specific or vague and ill-defined. This paper describes how formal interaction models may be used to increase the precision of these concepts with less dependence on implementation. An interaction model simply distinguishing display and state is introduced as a basis for the description of display editors. This model is then refined, first to incorporate notions of direct manipulation and then to model properties of pointers and boundaries (for use in cut and paste for example). In each case principles are formulated and related to previous work in the area.
Toward a Structured Approach to Specifying User Interface Design BIBA 415-421
  J. Preece; M. Woodman; D. C. Ince; R. Griffiths; G. Davies
Given the complexity of interface design it is hardly surprising that there are few published accounts of complete interface design methodologies. This paper describes how decisions from empirically collected data can be embedded in an overall decision-making framework.
Specifying the Interface Logic BIBA 423-428
  P. Jeremaes
The successful development of an interactive human-computer interface is dependant upon having a clear understanding of the structural properties that the interaction (dialogue) should possess. In this paper we present a logical framework for the specification of dialogue components. This framework distinguishes between explicit sequencing information and the implicit definition of relationships between dialogue units. The concept of a dialogue game is introduced to characterize a logic of interaction. Properties of this logic can then be used to aid analysis of the interface component.

2. Design and Evaluation Methods: 2.5 Dialogue Design and Evaluation

PAC, an Object-Oriented Model for Dialog Design BIBA 431-436
  Joelle Coutaz
PAC is an implementation model that attempts to bridge the gap between the abstract sphere of theoretical models and the practical affairs of building user interfaces. It takes as a basis the vertical decomposition of human-computer interaction into semantic, syntactic and pragmatic layers as promoted by some theoretical models. However, PAC stresses the fact that these notions do not form strict monolithic layers but are distributed across related "chunks", called interactive objects. For doing so, PAC recursively structures an interactive application in three parts: the Presentation, the Abstraction and the Control. The Presentation defines the concrete syntax of the application whereas the Abstraction corresponds to the semantics. The Control maintains the mapping and the consistency between the abstract entities and their presentation to the user. The Presentation of an application is in turn decomposed into a set of interactive objects, entities specialized in man-machine communication. As for applications, an interactive object is organized according to the PAC model. PAC has been used for the construction of two interactive applications and is currently applied to the development of a User Interface Management System.
WindowNet -- A Formal Notation for Window-Based User Interfaces BIBA 437-442
  Matthias Grochtmann
The dialogue description Language WindowNet is presented, which allows the description of complex, window-oriented dialogues. WindowNet is based on state transition nets, but has been enriched with variables and recursive parameterizable refinements. The functions of the window manager TUWin are integrated into the language. WindowNet automates the transition from functional specification to modular design and implementation.
Design Process and Operator Tasks During Automation of a Sugar Factory BIBA 443-452
  Alfred A. F. Brouwers; Frank D. Pot
The technological and economic determination of work organisation and job content has been questioned over the last years. Recent research focuses more on design processes and decision making
   This article is an analysis of choice opportunities regarding technology and work organisation during the automation of a beet sugar production plant. Little attention has been paid by the designers to problems of work and work organisation. The resulting operator tasks in two control rooms differ as a consequence of different levels of automation of the production processes that are controlled from these rooms. On the highest level of automation (computer-controlled processes) the complexity of the tasks is far less as compared to the lower level. This can be considered as degradation. The influence of operators, middle management and worker representatives was minimal during the design process and in decision making.
A Taxonomy of Evaluation Techniques for HCI BIBA 453-459
  Steve Howard; Dianne M. Murray
The results of an extensive literature search are presented and a rationale given for the existence of a number of diverse and distinct techniques used to evaluate the human-computer interface. A prescriptive taxonomy of the evaluation techniques uncovered in the literature search is proposed, and the emergence of an 'evaluation environment' as a necessary part of the taxonomy is described. Although much work needs to be done, the environment represents a first pass at a tool for matching techniques to requirements.
The Design and Evaluation of Online Help for UNIX Emacs: Access Mechanisms BIBA 461-466
  James E. Palmer; Thomas M. Duffy; Kathleen Gomoll; Thomas Gomoll; Jessica Palmquist-Richards; John A. Trumble
We discuss the use of the card sort technique and cluster analysis for determining an effective organization for a help menu in Unix EMACS. We gathered similarity data using a card sorting task with EMACS commands. We then used hierarchical cluster analysis to analyze the data. The results indicate that differences among novices, intermediates, and experts appear with computer-based concepts such as windows and buffers, but that the sorts are more similar than they are different. We argue that cluster analysis may aid designers in determining a functional organization, but that in our domain, this organization will not help users bridge the mapping from real world tasks to computer tasks.

2. Design and Evaluation Methods: 2.6 Participative Design

The Use of Participative Exercises in Human Factors for Education and Design BIBA 469-472
  D. F. Poulson; C. A. Johnson; J. Moulding
The paper describes participative techniques using group exercises, which have been used successfully as part of a training course in Human Factors for the designers and managers of computer based systems. The application of these techniques to the design process is also discussed, with an example of how this has been achieved in a practical application.
The Social Impact of User Models BIBA 473-478
  John M. Pratt
An interactive system may include a user model designed for a particular user. Such user models may include measurements of the personality of the user and the user then exhibits objective and subjective improvement compared with a standard environment. The personal nature of the model requires that it be kept confidential and secure, in both the private and public interest, and therefore action should be taken to ensure such protection.
Evolutionary Prototyping and the Human-Computer Interface BIBA 479-484
  S. Hekmatpour; D. C. Ince
The proponents of evolutionary software development regard a software system as a continually evolving entity rather than as the product of a series of developmental phases followed by a maintenance phase. This paper describes some of the human-computer interface components of an evolutionary prototyping system which is based on this philosophy. This system is based on a wide spectrum language which can support all the phases of the software development process.

3. Human-Computer Interface Design: 3.1 Human-Computer Interface Design

Evaluating User Interface Complexity BIBA 489-495
  John Karat; Richard Fowler; Mary Gravelle
This paper presents an attempt to utilize a formal model in the study of user interface development. A study was conducted to examine learning and performance differences between a command language and a direct manipulation system. Subjects initially unfamiliar with computer systems learned file management functions and carried out a series of tasks on one of the systems. Experimental results point out large differences in performance between the command language and direct manipulation systems which favor direct manipulation. Formal models of the knowledge required to use the systems were developed following the framework suggested by Kieras and Polson [1]. There are difficulties in mapping predictions from the formal models to the experimental data for the systems involved. Analysis suggests that inability of the formal model to predict error data was a basic problem with the formal analysis.
Are 'Programming Plans' Psychologically Real -- Outside Pascal? BIBA 497-503
  D. J. Gilmore; T. R. G. Green
The program plan (Spohrer et al, 1985) has been widely accepted as a description of programming knowledge. This paper presents an empirical test of their psychological reality and the first comparison of plan structures across languages.
   Subjects with 2-3 years' programming experience attempted to find deliberately introduced bugs of four types, one of which was bugs in plans. Graphic cues were used to highlight plan structures and/or control structures in a fully crossed design. Highlighting control structures improved the detection of control bugs in both Pascal and Basic, but only Pascal programmers benefitted from the highlighting of plan structures, showing an increase in the detection of plan bugs.
   We conclude that the 'plan' is a good description of Pascal programming knowledge. Can it be that Basic programmers do not use plans? That would be surprising because the structure of the programs was the same. We suggest that Basic programmers may use plans, but that the Basic notation is less "role-expressive", making it harder to identify plans and process them mentally. Research on other programming languages (eg. Prolog) is urgently needed if we are to confirm the plan as psychologically real.
Workshop Programming of Numerical Controls BIB 505-511
  G. Pritschow; R. Viefhaus
Which Task in Which Representation on What Kind of Interface BIBA 513-518
  Monika Gerstendorfer; Gabriele Rohr
For human-computer interface design it seems important to know more about the characteristics of tasks. The characteristics of a task have implications for how information should be represented on a user-interface. This would especially help in the discussion about designing graphical interfaces with direct manipulation or with natural language command structure and query. The lack of clear concepts of task parameters at this time makes the discussion difficult. An approach for determining task characteristics will be presented and first empirical proofs will be discussed with respect to consequences for user-interface design.

3. Human-Computer Interface Design: 3.2 Interface Design of Application Programmes

Report Generation Using a Visual Programming Interface BIBA 521-528
  Tim Dudley
Command language interfaces are not always the most appropriate tool at the initial stages of report design. A loosely constrained graphical notation can be much more useful. Visual programming techniques introduced on the Xerox Star, and popularized by the Apple Lisa and the Macintosh, have now made the use of such a graphical notation much more feasible. Also, the direct manipulation techniques described by Shneiderman are now viable because of the wide-spread availability of bit-mapped graphics screens and pointing devices such as the mouse.
   This paper briefly discusses visual programming concepts, and then describes the implementation of a visual programming interface (VPI) for a 4GL report writer. The basis for the design is an object-action syntax. A set of icons was designed which represent atomic report entities, and a graphic editor built to manipulate these entities into a report structure. Attribute sheets associated with each of the report entities allow definition of the report entities to the data dictionary. A menu bar controls menus of all possible actions to be performed on the objects. A facility to switch easily between the graphical and textual representation of the report is provided, with direct manipulation editing available in both representations. Modifications made in one representation are automatically reflected in the other.
   The combination of the VPI with a 4GL makes the design and modification of reports remarkably straightforward, and suitable both for end users and application programmers.
An Object Oriented Extension Language for Integrating Disparate Applications BIBA 529-533
  Kee Hinckley
Much of the work in user-interface design has concentrated on developing a consistent and friendly interface to particular applications. While this is appropriate in some environments, a broader problem concerns unifying interfaces to multiple applications, often from different vendors, whose interface designers were not working in concert. In a multi-window/multi-processing system, the user may be presented with different programs utilizing different user paradigms and requiring different methods to perform similar tasks.
   QUICHE (Quick User Interface and Command Handling Extension) is an extension language based on the Icon [1] programming language. Like Lisp-style EMACS [2] extension languages, it can be used to define key bindings. Unlike those extensions, QUICHE is not bound to any particular application, can easily make system calls and invoke programs, and can call entry points in the program it is extending, either directly or through a trait binding mechanism. It is this latter ability that makes QUICHE an ideal means of unifying application interfaces.
   Traits are defined for applications such as window managers, debuggers and editors. Each trait encapsulates a set of basic operations that will be available in any instance of an object. The existence of a standard set of traits allows the interface designer to tailor interfaces that provide consistency across multiple applications, and to easily modify or add features to a set of applications. In addition, QUICHE can also extend applications which do not themselves support any traits. A manager can be written in QUICHE that, instead of calling the application, inserts the commands corresponding to a call as though the user had entered them.
Human Computer Interaction -- A Framework for Analysis BIBA 535-540
  J. Kirakowski; J. Good
This paper presents a framework within which the dialogue between computer and human can be described. It is specifically designed to assist the psychological investigation of users' responses to adaptable and adaptive interfaces. As a descriptive tool it is theoretically neutral, and makes no assumptions about user intention or performance. However, it does have prescriptive implications for the design of interactive systems.
   A unit of human-computer communication is defined, and thence a way of characterising this unit from the point of view of the system being interfaced with; the actions open to the user; and the linkage between these characterisations. This linkage can be considered as a transformation of user actions to system directives. The paper argues that understanding this linkage, which up to now has largely been ignored or taking for granted, is crucial to our understanding of the psychological effects of an "adaptable interface".
   The paper presents a brief comparison with other interface construction methodologies and tools, and clarifies a characterisation of the system modularity which will be necessary in order to implement what are known as 'intelligent front ends' or resident knowledge-based assistants and help facilities.
Designing a User-Oriented Interface to a Document Management System BIBA 541-546
  A. Lesniewski; H. Rossler; P. Szabo; K.-H. Jerke
The significance of human computer interfaces with respect to the usefulness and effectiveness of a system is widely accepted. The most obvious directions used in building acceptable interfaces are based on the concept of object orientation and direct manipulation of visually presented objects. However the presentation of, and access to visual objects as well as performing actions on them can be crucial to the interaction style. In this paper an attempt is made to introduce some technical terms and to use them in formulating an appropriate conceptual model. The aim is to explore and evaluate the advantages of such direct manipulation.

3. Human-Computer Interface Design: 3.3 Natural Language Dialogues

The Man-Machine Interface: The Natural Language Barrier BIBA 549-554
  J. E. Grace
With the increasing use of computers in diverse applications it is becoming apparent that a major consideration in system design must be the methods of interaction between user and machine, especially when the user may not be formally trained as a Computer Scientist. The ideal situation would be the use of a Natural Language to communicate directly with the computer.
   Although the ultimate goal of Natural Language communication between user and computer would be via the spoken word, this paper will not consider speech comprehension and its associated problems, but will focus on written language and more specifically written English, although the problems and approaches referred to here are relevant to the speech problem and may be adapted to languages apart from English.
   This paper will discuss some of the systems developed to process Natural Language text, and will proceed to examine the major stumbling blocks to be overcome: lexical analysis, dealing with idioms, syntax analysis, semantic analysis, resolution of global and local ambiguity, processing pronoun referents, ellipsis, and the need for the construction and maintenance of a body or world and "general" knowledge to allow contextual understanding and pragmatic analysis.
The Natural Language Metaphor: An Approach to Avoid Misleading Expectations BIBA 555-560
  Andreas Kohl; Walter Rupietta
Natural language is often used as a query language for data bases. Up to now only subsets with restricted syntax and vocabulary in definite domains are implemented (often as prototypes). In spite of such restrictions, these query interfaces create a "natural language metaphor", which supports, but may mislead users.
   To overcome these problems, we investigate whether we may preserve the natural language metaphor, but prevent the user of misapplying it. Our approach is to specify the global metaphor in a set of more exact analogies and to supplement natural language input by other interaction techniques such as menus or direct manipulation.
   We do not stress the problems of natural language processing, but concentrate on the relevant metaphors. Our investigation is connected with the development of a natural language query prototype system for knowledge bases in the office domain.
What Do Users Say to their Natural Language Interface? BIBA 561-566
  William C. Ogden; Ann Sorknes
A controlled laboratory study was conducted to evaluate a natural language database query program. Our research goal was to assess how well a commercially available natural language interface would meet the needs of a database questioner who had no formal query training. The interface was evaluated by observing seven subjects as they learned and used the product. Results show that the interface had a difficult time understanding the subjects. Only 28% of their first questions resulted in a correct result and 16% of the problems were thought to have been correctly answered by the subject but in fact had not. These results are comparable to results of evaluations of other natural language products and suggest that natural-language-interface users will need to be trained.
Language Acquisition by Example in a Natural Language Consultation System BIBA 567-570
  Thomas Knopik
One of the difficult problems in expanding knowledge-based consultation systems is in specifying and entering the necessary linguistic knowledge for the natural language interface. A new approach to a part of this problem is the acquisition of lexical data for verbs by examples; i.e. by natural-language questions similar to those asked by the user during the consultation dialogue.

3. Human-Computer Interface Design: 3.4 Evaluation of Input Devices

The Effects of Various Types of Speech Output on Listener Comprehension Rates BIBA 573-579
  T. Moody; M. Joost; R. Rodman
This paper reports the results of a study that investigated the effects of varying speech signals, passages, and questions on listener comprehension rates. The passages and comprehension questions used were taken from sample tests of the High School Equivalency Examination, the Scholastic Aptitude Test, and the Graduate Record Examination. These passages were then recorded using four speech output types: synthesized speech, digitized speech at a rate of 9600 bps, digitized speech at a rate of 2400 bps, and natural (human) speech. A reading group was also used in this study for control purposes. Results indicated statistically significant differences in comprehension rates between the natural speech group and the synthesized and 2400 bps digitized speech groups. Significant passage and question type effects were also found. These results and the voice output guidelines derived from this study are discussed in this paper.
Optimal Size and Spacing of Touch Screen Input Areas BIBA 581-585
  Russell A. Benel; Brian C. Stanton
Touch sensitive displays have been implemented commonly as the simple interface to menu systems in information kiosks, but have the potential for use as the main interface to complex systems. The human factors necessary to design an optimal touch sensitive human computer interface have not been derived from an extensive base of empirical research. The available standards appear to have been developed originally for mechanical pushbutton switches. Two experiments were designed to evaluate the adequacy of the current standards for defining the size and spacing of the touch areas. The first experiment employed a Fitts' Law paradigm to evaluate distance, size and touch accuracy. The second employed a telephone-type touch entry keypad allowing input speed and accuracy to be evaluated with the differing size and spacing. Active touch areas were manipulated in conjunction with the variations in size and spacing. The results of these experiments are compared to the existing standards and the validity of the current standards are discussed.
Which Input Device Should be Used with Interactive Video? BIBA 587-592
  Cathy Thomas; Steve Milan
Two experiments and a literature review were carried out to attempt to establish which devices were optimum for entering information into a computer when using an interactive video system. Subjects performed different types of task using a number of different input devices. Measures of time, errors and subjective preference were used for evaluation. The results indicated that devices involving direct manipulation interfaces are superior.
Gestures as a Means for the Blind to Interact with a Computer BIBA 593-595
  Gerhard Weber
A new input channel in man-computer communication especially for blind computer users is opened. Fingers are used to form gestures on a touch sensitive input device. An implementation in a computer-aided dialogue to recognize gestures is described.

3. Human-Computer Interface Design: 3.5 Pictorial Information Presentation

An Investigation into Business Information Presentation at Human-Computer Interfaces BIBA 599-604
  Paul Booth; Chris Fowler; Linda Macaulay
Many commercial software packages aimed at the business community advertise graphical data presentation as a selling point. However, the general growth in the use of information display techniques in I.T. systems has not been matched by a proper consideration of their effects and at the moment, the evidence regarding display format preference is somewhat contradictory. An experiment is reported which was concerned with information display format (i.e: graphical and tabular) preference at the human-computer interface. Subjects from both business and non-business backgrounds performed a decision-making task after completing the visualiser-verbaliser and conceptual tempo cognitive style tests. The results suggested that the visualiser-verbaliser and conceptual tempo cognitive style dimensions are unlikely to be of use in predicting display format preference at the human-computer interface. Moreover, there were indications that individuals will use information in both graphical and tabular formats regardless of their own stated preferences. Nevertheless, as stated display format preferences were widely dispersed, it is suggested that users of computer systems should be allowed to choose the format in which information is presented.
Pictorial Communication with Computers BIBA 605-609
  Philip Barker; Mohsen Najah; Karim Manji
Human-computer interaction involves the movement of information between a human and a computer by means of suitably designed interface systems. Conventional interfaces for the transmission of text and other basic forms of data are now well established. Increasingly, various types of pictorial interface are being used to fabricate 'user friendly' dialogues with computers. This paper describes some approaches to human-computer communication via the use of conventional paper-based pictorial forms. Some attempts at evaluating end-user reactions to the use of this type of interface are described.
The Use of Structural Displays to Facilitate Learning BIBA 611-616
  L. Fitzgibbon; J. Patrick
Whilst there is evidence concerning the effect of various types of display on performance in human-computer interaction, there is little similar research investigating effects upon learning. This paper examines how the learner might be provided with displays which develop appropriate mental models prior to or during the learning process.
   Two experiments are described which investigate the effect of providing the learner with a display of the functional and procedural structure of a computer-based text editing task. In Experiment 1 this structural display is presented either prior to or after a series of learning modules concerning the criterion editing task. The results indicate that the structural display is of benefit generally, and more so when presented in advance of the learning material, as indicated by a reduction in errors and time to complete the criterion task. The same effect is found in Experiment 2 using a more complex editing task. The facilitating effect of the structural display is enhanced further when it is also available during the learning process. Generally the structural display appears to influence the encoding or assimilation of information which has to be learned.
Optimal Organizations Guided by Cognitive Networks and Verified by Eyemovement Analyses BIBA 617-622
  Kenneth R. Paap; Ronald W. Noel; James E. McDonald; Renate J. Roske-Hofstrand
Lee and MacGregor (1985) presented an analysis of search time that suggests that the optimal number of options per menu panel, over a wide range of conditions, is in the range of only four to eight. We have extended their analysis to cases when the options on each menu panel can be organized into meaningful categories (Paap & Roske-Hofstrand, 1986). Under the assumption that users can restrict their search to the relevant category, the extended model predicts optimal sizes that are much larger: 16 to 78 options per panel. Organizing items into categories will produce faster search times than alphabetical orderings (McDonald, Stone, & Liebelt, 1983). However, the magnitude of the benefit is likely to depend upon the degree to which organization of the database corresponds to the user's conceptual organization. For the past several years we have been developing a formal method for knowledge elicitation and representation that derives cognitive networks from rating and/or sorting judgments and applies these cognitive organizations to the design of human-computer interfaces. In this paper we show how the cognitive networks, in conjunction with a Category Quality metric, can be used to organize the options for a menu-driven interface. The method is validated through an eye movement analysis that permits overall search time to be decomposed into several processing components.

3. Human-Computer Interface Design: 3.6 Graphics in Human-Computer Interaction

Attraction and Distraction by Text Colours on Displays BIBA 625-630
  F. L. van Nes; J. F. Juola; R. J. A. M. Moonen
The effect of colour differences on visual searching in videotex displays has been investigated in several experiments, including one with accurate measurements of eye movements.
   Subjects had to search for specific target words on display pages with normal text in one, two or four colours. The target-word colour was either known or unknown to the subjects. Objective and subjective data from the experiments show the existence of two basic visual search modes in tasks of this type. The results of the reported research show the extent to which colour may attract or, alternatively, distract the reader's eyes while he/she is reading or searching a text.
   Some practical conclusions are drawn for information presentation on multicolour text displays.
Graphical Tools for Description of Dynamic Models -- Problems and Prospects BIBA 631-636
  Mikael Kindborg; Anita Kollerbaur
The use of graphical interfaces has given a large group of non-computer experts access to a powerful tool. Graphics have also been used in interactive environments that aid non-programmers in describing dynamic models
   Most existing graphical programming systems employ graphical formalisms developed within computer science. This is unfortunate because representations like flowcharts, state diagrams, constraint networks etc, tend to be hard to understand for non-specialists.
   In this paper we will examine some of the problems of graphical programming for users who are not computer specialists. Further, we will discuss how dynamic information is communicated in well known media like books, comics and film. Of particular interest is the static representation of time in comics. Finally, a novel approach to the design of pictorial simulation languages based on comics will be presented and discussed.
User Interfaces to a Medical Archiving and Communication System BIBA 637-642
  Manfred Ludtke; Ivar Nackunstz
The design of Viewing Stations emerging in digital radiology has to take account of the difficulties of non-technical users, pressure from routine work, complex dependencies with other working places and the need for compatibility with older systems. Philips Medical Systems has undertaken laboratory studies comprising two Viewing Stations for a PACS environment, one being a rapid prototype, and a Radiology Information System. These studies address questions of interaction syntax and the conceptual model of a user. A desk/lightbox metaphor is successfully applied to the viewing station design.
An Abstract Model for Interactive Pictures BIBA 643-647
  Vincent Quint; Irene Vatton
This paper presents the user interface tool kit developed for Grif, an interactive system for structured document manipulation. We focus on those concepts which led to the design of this package and to their application in a document production system.
   The tool kit provides features which allow display of the visual aspect of documents and which handle user interaction. Most of these features are based on an abstract description of the pictures to be displayed and through which the user interacts with the application. Pictures are described in terms of boxes defined through position and dimension constraints. It is shown that this representation leads to device independence, user tailorability and a high level of interactivity.

3. Human-Computer Interface Design: 3.7 Learning and Training

Learning a Computer System by Unassisted Exploration BIBAK 651-656
  Jean-Marc Robert
The goal of this paper is to investigate the dynamics of exploration as a learning mode for computers and eventually improve system explorability. An experiment was conducted on learning a computer system by unassisted exploration. Five subjects with previous computer knowledge explored a system for two hours. Their verbal comments were recorded, and the session was videotaped and timestamped. The main results show that the exploratory process may be decomposed into a series of work episodes which in turn may be broken into chunks of behavioral and cognitive activity. A chunk is centered on some task and thus includes a goal, reasoning modes, a strategy, difficulties (frequently), and an outcome. This paper examines that way goals are set, the strategies and reasoning modes that are involved in exploration, and factors responsible for transitions from one chunk to another. Recommendations for improving the explorability of computer systems are proposed.
Keywords: Exploration, Learning, Learning by doing, User interface, Cognitive ergonomics, Human-computer interaction, Human factors
How to Use Plan Recognition to Improve the Abilities of the Intelligent Help System SINIX Consultant BIBA 657-662
  Matthias Hecking
On the average, users of complex computer systems (e.g. UNIX, SINIX, TEX) use only 40% of all functions available.
   In order to improve this situation, intelligent help systems offer the passive capability of responding to natural-language questions. But, this is only then sufficient if the user is familiar with the underlying concepts involved.
   To make sure the user is aware of these unknown concepts, the help system must be an active one. The system has to closely examine the typed-in commands in order to infer the underlying goals of the users non-verbal activities.
   In this paper, a brief preliminary survey of the intelligent help system SINIX Consultant (SC) is given, concentrating mainly on explaining the realized plan recognizer REPLIX and its employment within the SC system. Especially the following questions and the solutions are discussed:
  • how to model inserted sub-plans and overlapping plans,
  • how to treat commands which don't belong to any plan (ignore commands),
  • how to use interrupt commands for plan focusing, The paper concludes with the presentation of an extensive example.
  • Designing Systems for Training and Decision Aids: Cognitive Task Analysis as a Prerequisite BIBA 663-668
      Renan Samurcay; Janine Rogalski
    The general aim of this study is to define the characteristics of a system designed to aid users both in decision making and training, in situations which share certain properties with supervisory control tasks. It focuses on the Method for Tactical Reasoning (MTR), which is a guideline for defining and reaching an optimal target state in complex situations concerning public safety. The method is analyzed within the Rasmussen's framework. This theoretical examination suggests that the computer and the operator should act serially. Observations of safety officers during training in MTR provided data on mental strategies and required background knowledge. These finding are discussed in terms of the definition of properties of computer-based training systems.
    Trend Presentation on VDT as a Decision Aid to Operators BIBA 669-674
      Ted N. White; Pedro van der Meijden
    The paper discusses the results of five trend experiments. Different forms of trend presentation, based on changing the time and amplitude scale of the trend, are investigated. The prediction accuracy of the subjects is measured, statistically tested and thereafter compared with the ratings of the subjects perceived performance. Moreover, the prediction strategy is investigated.

    3. Human-Computer Interface Design: 3.8 Graphical Workstations

    An Interactive 3D-Graphics User Interface for Engineering Design BIBA 677-682
      Andreas Brunn; Klaus Lay; Uwe Rettich
    An interactive 3D-graphics user interface for the design and layout of production systems was developed at the Fraunhofer Institut fur Arbeitswirtschaft und Organisation (IAO). Qualities like uniformity, openness of the system and software ergonomical criteria like flexibility, good handling and userfriendliness played an important role during the design of the interface. The graphic manipulations are based on objects which are built up with primitives (for example cubes, cylinders, prisms) or which can be transmitted from CAD-systems. Different windows which show the actual layout on the screen can be defined, modified or deleted. The user can choose between 2D- and 3D-representation (hidden surface or wire-frame model). The movements of objects within one arrangement are shown in all corresponding views on the screen simultaneously. Menus are guiding the user within the system. Input facilities are the mouse and the keyboard. Help functions are providing additional information about graphic objects and about all available functions. An application of the 3D-graphics user interface is shown at the example of a robot simulation program.
    A Framework for Comparing Systems with Visual Interfaces BIBA 683-688
      Ted Selker; Cathy Wolf; Larry Koved
    A computer program presents its capabilities and domain of application through a user interface. With the advent of inexpensive graphics hardware, systems with visual user interfaces are proliferating. New interface technologies offer opportunities for improving the usability of programs. It is important to understand how to employ these new techniques in the design of better user interfaces. A review and comparison of visual interfaces prompted the need for a vocabulary and systematic framework to describe them. This paper presents a framework developed for describing and comparing visual user interfaces.
       Communication between computers and humans has often been described in linguistic terms. This paper uses the term visual language to refer to the systematic use of visual presentation methods to convey meaning to a user. The framework includes a description of the interface in terms of the elements, operators and syntax of an interface language, the rationale governing the use of visual elements, the power of the language, interface characteristics such as the interaction style and input/output device dependencies, and the domain and purpose of the application.
       The framework has been useful in identifying important differences between visual interfaces and has provided a vocabulary for the discussion of visual language. Elements of the framework are illustrated with examples from existing systems.
    Linking Multiple Program Views Using a Visual Cache BIBA 689-694
      Charles Wiecha; Max Henrion
    To reduce disorientation in Demos, a non-procedural decision support system, we have built a graphical interface to display models as a hierarchical set of node and arc diagrams. Through empirical studies on user interactions with the diagrams we develop the idea of a visual cache. A visual cache can improve access to information on a display screen by organizing multiple views to take advantage of locality in a user's patterns of information search and exploration. The notion of the visual cache may be a useful principle in designing graphic interfaces to avoid disorientation.
    A Structural Model for Hierarchically Describing Human-Computer Dialogue BIBA 695-700
      Deborah Hix; H. Rex Hartson
    A variety of "dialogue models" has sprung into existence over the last decade in response to the need for organizing the process of human-computer dialogue development. Structural models describe the generic process of human-computer interaction, and can be used to guide a dialogue developer in constructing the dialogue. The dialogue transaction model presented in this paper is such a structural, descriptive model. It identifies linguistic objects in the behavioral domain, and defines linguistic processing of those objects in the constructional domain of the dialogue. This paper presents the model, how it was derived, and how it is used to describe human-computer dialogues.

    3. Human-Computer Interface Design: 3.9 User Differentials

    Multi-Level User Interfaces: Software Tools and an Application BIBA 703-708
      J. Kraak
    Multi-level user interfaces are appropriate for systems with different types of users. Differences in system knowledge and task structure are of great importance to the design of software tools for these interfaces. The general purpose toolkit DIALOOG supports the implementation of common alphanumeric dialog types, suited to beginners as well as experts. Interfaces for different task structures can be made by the User Interface Management System COMMAND. The three-level interface of the graphing system KOMPLOT illustrates the multi-level interface approach, which makes interactive software accessible to a wide range of users.
    Gender and Cognitive Style Differences at the Human-Computer Interface BIBA 709-714
      C. J. H. Fowler; D. Murray
    Over the past decade the computer user community has become increasingly heterogeneous. Computers are now much more widely available and are to be found in most schools and places of work. In response to the greater variety of user, designers have become more concerned with the issue of usability. This has led to a growing need for a fuller understanding of the significant differences between users and user groups so that attempts can be made to accommodate these differences in the design of future systems. Two possibly important sources of variation between users are sex and cognitive style. This paper aims to explore these two concepts, to discuss some of their implications for system design and to finally highlight those aspects which may require further investigation.
    Levels of Adaptivity in Interface Design BIBA 715-722
      P. A. Totterdell; M. A. Norman; D. P. Browne
    The term adaptation has been used variously to describe computer systems which are personalisable by either designer or user, systems which fixedly change in response to user or application features and to systems which assess the value of their changes. This paper attempts to identify a coherent framework into which these seemingly disparate systems can be mapped. The foundation for the framework originates from the observation that adaptation relates a system to its environment. A number of levels of adaptivity are revealed by using a model of the computational strategies which can be applied in playing game theory's prisoners dilemma. Sufficiency and boundedness are introduced as key concepts. The levels are reinterpreted by looking at an account of adaptation in evolution. The adaptivity argument is then discussed in the context of explanation types in science and in particular Bechtel's realistic-intentional approach. Some design methodology principles for adaptive systems are suggested with Dennett's notional worlds as a first step in the design. Finally, the paper identifies a taxonomy for adaptive systems and reviews the state of the art in computer adaptive systems with respect to the taxonomy.
    Adaptability and Tailorability in NoteCards BIBA 723-728
      Randall H. Trigg; Thomas P. Moran; Frank G. Halasz
    NoteCards is an information structuring system developed in the Intelligent Systems Lab at Xerox PARC. A major design goal has been that NoteCards be an adaptable system, that is, tunable or customizable by users for particular applications and styles of use. In this paper, we describe four ways that a system can be adaptable: (1) it can have a flexible underlying conceptual model, (2) its behavior can be parametrized, (3) it can be integratable with other facilities, and (4) it can be tailorable, i.e. users themselves can add new functionality. We discuss the adaptability of NoteCards according to each of the above criteria. Finally, an example of large scale tailoring in NoteCards is presented.

    3. Human-Computer Interface Design: 3.10 Techniques of Dialogue Design

    Structured Command Interaction Based on a Grammar Interpreting Synthesizer BIBAK 731-737
      Sten Minor
    This paper describes a generic user interface for command interaction based on a grammar interpreting synthesizer. The interpretive approach facilitates interactive modifications of the command grammar which is useful both for design and personalization of the command interface for a specific application.
       A brief description of the underlying synthesizer is given. The command grammars are described and the consequences for the interaction are examined. Two approaches for integrating the command interface with an application are given together with a discussion of the applicability of each approach. An example of the interaction style in a command synthesizer is given and the relation with other styles of interaction is discussed. The possibility to personalize the interface by modifying the grammars is examined and some directions for future work are given.
    Keywords: Software engineering, Tools and techniques, User interfaces, Human-computer interaction, Interaction techniques, Command grammars, Adaptable user interfaces, Synthesizers
    Executable Specifications as an Aid to Dialogue Design BIBA 739-744
      Heather Alexander
    Designing the user interface is still an experimental craft, primarily because it is difficult to be precise and objective about what constitutes a "good" design. Consequently, interface designers need to be able to experiment with different ideas, subjecting those ideas to user evaluation at an early stage in the process. This paper uses techniques from software engineering to formalise and prototype user interface designs.
    Dialogue Issues for Interactive Recovery -- An Object-Oriented Framework BIBA 745-750
      Martin Rathke
    This paper describes an application independent framework for interactive recovery in computer-based dialogues using an object-oriented programming language. The implemented system allows an application programmer to specify methods to save system states efficiently and to return to prior system states. System states may be defined for single objects or groups of objects or for the total world of the underlying application. Transitions between system states may be defined in a state-oriented manner, i.e. by changes of values of state-variables, or in a function-oriented manner, i.e. by transition functions. All system states may be recovered and analyzed with the support of a visualization tool in a window-based system.
       The advantage for the application programmer using this framework is that he only has to specify descriptions of methods for recovery (a classification of undo-methods) but doesn't actually have to implement algorithms to restore prior system states.
    IMAGES -- An Object Oriented UIMS BIBA 751-756
      Luis Pinto Simoes; Jose Alves Marques
    This paper presents the architecture and functionality of an object oriented UIMS for a multimedia workstation which is being developed on the SOMIW project (Esprit 367). The internal architecture of the UIMS is described along with the functionality of the main components. A two level hierarchic control was adopted, where each application is managed by a delegate UIMS that communicates with the central UIMS. Applications, according to the object-oriented paradigm, are composed of several cooperating objects which implement the lexical and syntactical level of the dialogue. Finally the interface generator is described. It provides the applications programmer with a simple tool for specifying and prototyping the interaction with the user.

    3. Human-Computer Interface Design: 3.11 Display Systems

    An Approach Towards a Truly High-Level and Integrated User-Computer Interface BIBA 759-764
      Dietrich W. Paul; Hans R. Wiehle
    A further raising of the conceptual and linguistic level of the user-computer interface is suggested. This allows, to a high degree, the incorporation of the user's technical terminologies. The necessity of both consistent modes of working and of ready-to-use environments for the user's various tasks yields and demands an architectural model of the complete user-computer interface. This finally requires advanced cognitive models of the user and intense efforts at standardization. This paper holds that these challenges must be embraced urgently and demonstrates how this could be done. Sect. 1 presents this approach and its embedding into the setting of the actual situation of application oriented data processing. Sect. 2 presents a worked example in user oriented high-level problem solving. Sect. 3 outlines basic requirements and concepts towards such an architectural model.
    Human Factors Research in Remote Display Systems BIBA 765-770
      Denise C. R. Benel; Walter T. Talley; John D. Addington
    The U.S. Postal Service processes millions of mail pieces daily using sorting machines that have poor ergonomic design, suboptimal throughput, and associated operator fatigue. Two experiments are discussed on display imaging requirements conducted as part of the research leading to the envisioned new generation of sorting machines that will digitally capture a mail piece image, route it to an operator who views the image on a CRT screen and keys in the destination code which controls the sorting of the mail piece.
    Viewing Geometry of Single or Multiple Screen Displays with Planar or Curved Surface BIBA 771-776
      Georg Geiser
    The viewing geometry of single or multiple screen display systems determines the observer area and observer space. Screens with convex curvature have considerably smaller observer area, whereas concave screens have a substantially increased observer area compared with the planar screen. Furthermore concave screens show a more uniform brightness with less annoyance due to reflection. Besides rules for the arrangement of screens, it is concluded that the feasibility of concave screens is to be investigated.
    Some Theses on Undo/Redo Commands BIBA 777-781
      Klemens Waldhor
    UNDO/REDO are important components of software systems which claim to be easy to use. Different aspects and requirements such as different kinds of users are discussed. It will be shown that an UNDO/REDO component should contain more than only enable the user to undo his operations. The user needs sophisticated support to work easy and time effective with such a component.

    4. Impact of Computers on Human Behavior: 4.1 Impact of Computers on Human Behavior

    Technology Assessment Concerning Impacts of Information Systems BIBA 787-792
      Hans-J. Bullinger; Klaus Kornwachs
    Technology Assessment has become an important tool to describe, predict and to anticipate the possible control of the development of information techniques. Particularly the design, structure and potential use of Expert Systems has given rise to an impact on social, organizational and technological level. The aspects of users competence and future developments of man-machine interfaces are discussed and some consequences are shown.
    Getting the Baby into the Bathwater: Putting Organizational Planning into the Systems Design Process BIBA 793-797
      Susan M. Dray
    It is a truism that "people problems" are the major determinants of system success or failure. However, planning for the human and organizational changes which accompany technical change is still more the exception than the rule in many places. We have been working to change that at Honeywell since 1984, when we were first commissioned by our CEO to look at the impact of technology on the Corporation's culture. Since that time, we have been deeply involved designing a way to modify the traditional systems development process in order to incorporate proactive planning for human and organizational impacts of new information technology. The result is a methodology called DELTA (for Delivering Education to Leverage Technology Application). This paper discusses the way in which we designed this new methodology and will describe it, giving examples of its use from the consulting we have done.
       DELTA is based on the assumption that to be fully successful, technology implementation requires input from three perspectives: Business, Technical and Organizational. A major benefit has been the increased cooperation between professional communities which previously were unaccustomed to working together. By taking a systemic approach to technical system development, we feel that it is far more likely that human and organizational issues will be addressed proactively.
    Factory Workers and the Language Barrier - Making Computers a Tool Rather than a Nuisance BIBA 799-803
      H. Weule; L. Loffler; Th. Selinger
    In a field, where efficiency can be measured in terms of profit, user acceptance is of vital importance when it comes to introducing new systems. Too often, a new system fails or proves too costly to be desired when it takes special skill to communicate with the computer.
       Using a computer by non-professionals, such as shop workers in a factory, may call for a wildly different philosophy for the human interface.
       One one hand, it cannot be expected from the workers to undergo special training in the basics of data processing. In a time where hardware and software are making giant progress it should much rather be expected that computers adjust themselves to the level of their user. On the other hand, the degree of efficiency gained when really putting computers to work on the factory floor, could be immense.
       In this paper, the programming of numerically controlled (NC) tool machines is taken as an example to discuss a number of aspects and possible improvements.
    Social Evaluation of the User Interface: Who Does the Work and Who Gets the Benefit? BIBA 805-811
      Jonathan Grudin
    When an application requires the involvement of several users, evaluating its functionality and interface becomes more complex: that which benefits one user might not benefit another. An application program written to support cooperative work may present a systematic imbalance in the efforts required of and benefits obtained by different categories of user. Such imbalances may affect the acceptance and use of a product in unforeseen ways. The collective benefit to the group may be difficult to measure, and even if established, may be difficult to communicate effectively to those who do not benefit directly. In weighing a potential development project, decision-makers may be inordinately influenced by the attractiveness of the system to managers such as themselves, and not perceive that the requisite cooperation of other users of the application will not be forthcoming when those users do not benefit equally. In the absence of careful analysis, decisions to build unworkable systems are not only possible, but likely. These points are elaborated through the examination of several multi-user application areas in the context of evolving technology trends, organizational practices, and social tendencies.

    4. Impact of Computers on Human Behavior: 4.2 Organizational Issues of Computer Use

    New Technology and Job Satisfaction -- A Case Study of Travel Agents BIBA 815-820
      Howard Kahn
    Reports field research to determine whether a) the amount of time spent using a computer system and b) the amount of the job the computer system can do, are related to the user's perceptions of job satisfaction and the perceived job characteristics. The Hackman-Oldham Job Diagnostic Survey was used with 185 Travel Trade employees. Results suggest that perceived job satisfaction increases while the job's motivating characteristics vary with the amount of time spent using the computer system; and that perceived job satisfaction is unchanged while the job's motivating characteristics vary with the amount of the job the computer system can do. A number of other results are noted, together with the implications for management.
    Comparative Factors in User Acceptance of Office Automation BIBA 821-825
      Joan M. Straub; John R. Straub
    In a published study conducted by the authors in the early 1970's, evidence was found that user acceptance of computers into an organization was dependent upon how the users perceived the probability of ultimate self-enhancement of their occupational status as a result of the new technology. Current studies by the authors based on companies installing new hardware systems and related software (office automation and personal computer systems) indicate that high acceptance for new technologies continues to correlate most positively with intangible rewards, including, but not limited to, perceptions of job enhancement and increased self esteem. Other phases of office automation involving user control of the physical environment (ergonomic workstations, etc.) become effective tools for acceptance when they too can be perceived by users as occupational role enhancements.
    Research on the Use of Computer Based Message Systems in Organizations -- The Swedish IDAK-Project BIBA 827-832
      Hans Kohler
    The multi-disciplinary project IDAK focused on the use of computer based message systems in some Swedish places of work. It showed that horizontal and diagonal contacts were expanded by the use of the message systems and that there is a potential for more trade union contact through the systems. An important general property of message systems is the power to make communication objective. This gives new possibilities for the understanding of organizations. It is possible to use message systems for organizational development, but there is a need for social learning and users must develop a new communicative competence. The collective time dimension can collapse and social dynamic processes can take over. One basic problem is to balance the power of the sender and the power of the receiver of any message.
    Automated Monitoring, Feedback, and Rewards: Effects on Workstation Operators' Performance, Satisfaction, and Stress BIBA 833-837
      Delbert M. Nebeker
    The effects of automated computer monitoring were examined in a simulated organization. Twenty-four computer workstation operators worked under various conditions over a nine week period. Individual keystrokes per hour, productive time and performance against standards were monitored by their computers. Feedback reports on performance and incentives earned were available on demand at their consoles. Effects of different levels of standard difficulty and incentives are reported for performance, satisfaction and stress.

    4. Impact of Computers on Human Behavior: 4.3 Novice Training and Learning

    From Novice to Expert User: A Transfer of Learning Experiment on Different Interaction Modes BIBA 841-846
      Norbert A. Streitz; Will A. C. Spijkers; Lidewij L. van Duren
    In a transfer of learning experiment (40 subjects in a 4 group between subjects design) the question was investigated whether in text-editing prior experience with menu selection affected the subsequent learning of a control command language in comparison with learning the command language from the beginning. A 2x2 factorial design was used where the factors were presence vs. absence of prior experience and type of help device (menus vs. a help window belonging to the command language).
       Experienced users can reach a high speed using the control command language. But for novices, who do not yet have available the necessary knowledge, interaction speed is very slow. They are better off with menu selection, because this dialogue mode requires hardly any memorization of commands. However, if those users have no reason to remember anything, do they have to start from "scratch" when they want to use the command language.
       From the four groups of novices two groups gained prior experience with either menu selection or the control command language by editing several texts. The other two groups did not take part in the initial interaction. This was followed by the learning part, in which texts were edited using the command language only, but with menus or the help window as help devices. For all groups acquisition curves for a limited set of codes were assessed, and they were tested on their overview over the whole range of available commands.
       It appeared that novices using menu selection gain a better notion of the total range of commands than novices using a command language. Contrary to an often made assumption that prior experience with menu selection interferes with subsequent learning of control commands, we found no negative or positive transfer.
    The Trouble with UNIX: Initial Learning and Experts' Strategies BIBA 847-854
      Anker Helms Jorgensen
    UNIX Mail is an electronic mail system in world-wide use. Like UNIX it is claimed to have serious shortcomings in terms of usability. This paper reports on an investigation of the learnability of UNIX Mail. In addition, differences in the learning strategies of computer scientists and clerical staff were investigated.
       Two studies were conducted: The first was restricted to basic UNIX Mail features while the second included advanced features as well as a few UNIX features. Fifteen subjects, unfamiliar with UNIX Mail, solved simple message handling tasks in UNIX Mail while "thinking-aloud".
       The users experienced difficulties in nearly all areas, e.g. underlying mail concepts, dialogue, e.g., lack of feedback and phrasing of messages, confounding of modes, scope rules, help facilities and documentation.
       The causes of the difficulties are discussed and general implications for design of electronic message systems are proposed. Complementing the investigation of learnability, two expert users of UNIX Mail were interviewed on their views on the usability of the system.
    Impact of Feedback Content in Initial Learning of an Office System BIBA 855-859
      Jean McKendree; John M. Carroll
    The ability to discriminate between correct and incorrect behavior is essential for learning to take place. This is usually accomplished via feedback from the environment. However, there has been little systematic work on principles of feedback in order to promote optimal acquisition of new skills. The study presented in this paper investigated different types of feedback content in the context of an on-line learning-by-doing tutorial for an office management system. The results suggest that blocking of errors by means of a "scenario machine" is a somewhat beneficial type of feedback, but that including positive content can make learning even more efficient. Feedback about the goals of tasks or the steps necessary to perform typical tasks were found to promote faster learning and better performance on tasks when informative feedback was withdrawn. Implications for design and analysis of interfaces are discussed.
    Conceptual Models in Training Novice Users BIBAK 861-867
      Maung K. Sein; Robert P. Bostrom; Lorne Olfman
    Research findings provide ample support for the notion that providing a conceptual model of a system aids novice users in building mental models. Two types of conceptual models have been proposed: Analogical models and abstract models. It is argued that abstract conceptual models will provide more flexible mental models because they are more flexible representations. Individual differences must be considered as moderating variables in this analysis. A laboratory experiment was conducted to compare analogical and abstract models. Graduate business students were trained in the use of the filing capabilities of an electronic mail system. The findings of this study indicate that abstract conceptual models may indeed provide more flexible mental models for novice users of electronic mail, and that individual differences (in this case, visual ability), do affect the formulation of mental models.
    Keywords: Abstract models, Analogical models, Conceptual models, Electronic mail, Individual differences, Mental models, Novice users, Software training, Visual ability

    4. Impact of Computers on Human Behavior: 4.4 User Needs

    Customizing Help Systems to Task Structures and User Needs BIBA 871-878
      Rainer Lutze
    A systematic approach is presented for deducing the necessary features of a help system and its implementation complexity from an analysis of the task to be solved by an application system and its anticipated user group. By an analysis of the task structure with respect to the user group, first the necessary help services to be offered are determined. This is done by attaching help services to points within the space of possible task structures spanned by action preparation and execution on one side and motivation and execution control on the other side. A classification of the user group will advise the depth of the necessary processing of help requests, focusing on information, consultation or recommendation services. From these two information sources, the corresponding software technology of the help system and the necessary constraints on the application system can be extracted. Help systems will be categorized with respect to their passive or active activation mode on one side, and their static or dynamic behavior on the other side. As a result, the costs of the help system and the chances for its realization in a short-term, medium-term or long-term time horizon can be roughly estimated.
    Computerizing Data Presentation and Analysis BIBA 879-884
      M. Corbett; J. Kirakowski
    Many aspects of the computerisation of tasks require serious consideration by the human factors specialist. Among them is the issue of the cognitive impact on the user of 'computerizing' a data presentation and analysis task.
       Askwall (1984) has demonstrated that global measures are not sufficiently sensitive to isolate differing patterns of performance. In her experiment overall measures of time were spent on different aspects of the task.
       The effects of computerization in the present paper are examined by comparing Ss performance on an interactive computer program with that of Ss who received a similarly informative series of displays on slides. As predicted from Askwall, users of the computer differed from those using slides in the way in which they approached the task. This further confirms that global performance measures may not indicate any differences but the differences are more likely to be found in the types of display that are used to obtain the solution. By content analysing Ss' verbal protocols, we can begin to understand what sorts of changes and shifts of emphasis are necessary in order to 'computerize' this sort of task effectively.
    The Role of the System Image in Intelligent User Assistance BIBA 885-890
      James R. Miller; William C. Hill; Jean McKendree; Michael E. J. Masson; Brad Blumenthal; Loren Terveen; Jay Zaback
    Many researchers have demonstrated the ways in which well-designed graphical interfaces allow users to acquire conceptual models of the interfaces and the application programs behind these interfaces. It is also clear that the users' models are initially flawed and incomplete, and the problems that users have with these systems revolve around the misconceptions and alternate conceptualizations in these models. Our work indicates that graphical interfaces may be especially sensitive to misconceptions, and that advisory systems for these kinds of systems must anticipate and resolve such problems. In particular, they must be able to understand alternative conceptual models of the system, and may need to diagnose and remediate these misconceptions. We will describe our work on direct manipulation interfaces and intelligent advisory systems, focusing on the problems people encounter when using interfaces and the ways in which our advisor's design is being driven by the properties of graphical interfaces.
    Transfer of Learning in the Real World BIBA 891-896
      Mary Beth Rosson; Nancy L. Grischkowsky
    Twenty-one users of two existing procedural formatting systems were studied during a week-long course on a new tag-based system. Users' performance on class exercises was monitored, and subjective reactions were assessed at several points in the process. Users were classified along two experience dimensions: formatting history, and intensity of use. Our analyses revealed that heavier users of the existing systems made more errors on the exercises; particularly common were omitted and extraneous commands. We also found that attitudes of the participants diverged over the five-day period as a function of formatting history, with the more sophisticated users becoming more negative, and the more casual users more positive.

    4. Impact of Computers on Human Behavior: 4.5 Cognitive Aspects of Information Retrieval

    Software for Reading Text on Screen BIB 899-904
      D. J. Pullinger; T. I. Maude; J. Parker
    A Humanised Interface to an Electronic Library BIBA 905-910
      I. D. Benest; G. Morgan; M. D. Smithurst
    This paper describes a library metaphor which closely models the mechanisms for conventional searching, acquiring and reading of paper documents and is based on the premise that models which already are well learnt will be easier to use when similarly presented on-line. The model for manipulating the library catalogues described in this paper is based upon a dictionary metaphor that is an extension of the more general book metaphor. The book metaphor provides an open book presentation, with animated page turning across the screen and all the functionality that would be provided by a real book. It is argued that this interface greatly eases the use of a computerised library as it is based upon established and intuitively obvious concepts, and implicitly conveys cues with which the human user is already familiar.
    Comparing Words and Icons as Cue Enrichers in an Information Retrieval Task BIBA 911-916
      M. W. Lansdale; M. Simpson; T. R. M. Stroud
    Cue enrichment is a process whereby computer based information is associated with additional cues which can be remembered and used in retrieval. In this respect, it has become particularly fashionable to consider the use of visiospatial information such as shapes, colours and locations. This study aimed to look at the memorability of these cues, and to make a formal comparison with keywords used in the same way. Both methods were studied under conditions in which the enriching cues were assigned automatically by the system, and in which the user selected them explicitly. Little difference was found between the verbal and visual cues in terms of their overall memorability, but interesting qualitative differences emerged, both between the visual and verbal modes and between the different dimensions of the visual and verbal cues.
    Advanced Organizers in Computer Instruction Manuals: Are They Effective? BIBA 917-921
      Barbee T. Mynatt; Katherine N. Macfarlane
    An important issue in writing computer instruction manuals is how to effectively present the semantic material. Research on instructional methods has suggested that advance organizers (e.g., metaphors) can be an effective tool in teaching certain types of material in some situations. In many cases, teaching computer skills satisfies these criteria. In our research, beginning programming students were trained on a programming concept. A session involved reading an instruction manual and entering lines of source code into an interactive system. One group was given a metaphorical advanced organizer prior to training. A control group read a passage on computer history prior to training. The groups did not differ on training task performance. However, the advanced organizer group did better overall on tests of near and far transfer, and did proportionately better on the far transfer test of their semantic knowledge. Factors relating to when to use an advanced organizer are discussed.

    4. Impact of Computers on Human Behavior: 4.6 User's Language

    Digressional vs. Semantic Subordination: On the Role of Menu Structure for Users' Understanding of a Human-Computer Dialogue BIBA 925-929
      Kerstin S. Eklundh
    The design of a human-computer dialogue must be based on a structural analysis of the activity in which the computer system is to be used. In this paper it is argued that one should distinguish between two basic types of operations in the consideration of embedding of operations in a human-computer dialogue, and thus e.g. in the design of menus: 1) "digressive" operations, 2) operations which are semantically subordinated with respect to the current activity. In many systems, these two types of operations are treated as equivalent and appear in the same menu. With reference to two earlier studies of menu-based interaction, and on the basis of some research on human-to-human dialogues, it is claimed that users' understanding of modes can be expected to be substantially different in these two types of embedding in a dialogue, and that digressive operations are the ones that should preferably be implemented in a "modeless" fashion.
    Who's the Boss: Talking to Your Computer in the AI Age BIBA 931-936
      Kenneth H. Abrams
    Software user interfaces that mimic humans can have problems of unrealistic expectation and command ambiguity. It is proposed that such systems adopt human roles with limited responsibility that will maintain user expectation at a level that can be satisfied, and that implicit requests be avoided.
    A Summary of Experimental Research on Command-Selection Aids BIBA 937-942
      Jay Elkerton; Robert C. Williges
    Two experiments were conducted to assess and improve novice performance in hierarchical file search. In the first experiment, half of the novices were performing like experienced subjects, using search procedures that exploited the structure of the file hierarchy for efficient information retrieval. Other slower novices used time consuming scrolling strategies and appeared unfamiliar with the file hierarchy. In the second experiment, command-selection aids improved the search performance and strategies of slow novices when the underlying command-selection models were not unduly complicated (i.e., frequency profiles and search plans) and did not introduce a highly interactive, mixed-initiative dialogue.
    Developing Exploratory Strategies in Training: The General Approach and a Specific Example for Manual Use BIBA 943-948
      Rigas Wendel; Michael Frese
    Exploration is an important factor in learning how to use a computer. In manual construction, exploratory behaviors can be supported (e.g., by modularity and task orientation) and necessitated (e.g., by presenting a random sequence of information modules). In an experiment with 21 subjects, it was shown that the principles of manual construction advanced here led to better performance in comparison to a commercial manual. The exploratory manual was successful in inducing exploratory behavior.

    5. Forefront Systems and Techniques: 5.1 From the User's Point of View

    GIOTTO, An Intelligent User-Assistance BIBA 953-957
      Grazia Butera; Francesco Pastore
    The application of AI techniques to the management of a documentation base is a relatively new domain for AI. This paper outlines the problem of simulating, as far as possible, the human approach when it needs to store or retrieve information according to the semantics, supplying an intelligent help to the user when he is looking for some information, but he don't know exactly what.
    Catalogues: A Metaphor for Computer Application Delivery BIBA 959-964
      Stuart K. Card; D. Austin, Jr. Henderson
    This paper presents the mail-order catalogue as a metaphor for the delivery of application software in an integrated work environment. It also describes, Catalogue, an adjunct to the Rooms multiple virtual workspace environment, which employs this metaphor. This mechanism can be used (1) to give users "instant starts" by letting the users select a standard setup, (2) to allow users to assemble their own environment from standard components, (3) to parameterize a standard component, and (4) to load applications ready to run.
    Volunteering Information -- Enhancing the Communication Capabilities of Knowledge-Based Systems BIBA 965-971
      Gerhard Fischer; Curt Stevens
    Cooperative problem solving systems support the solution of tasks which cannot be solved by the human or the computer alone. These systems need to be knowledge-based and require flexible communication paradigms allowing natural communication with both experts and novice users of the system. Natural communication (quite different from natural language) has to support mixed-initiative dialogues where information can be volunteered by the system and the user.
       In this paper, we present prototypical systems which assist users in rebooting a computer. REBOOTER is a rule-based system which guides the user with a strongly system-directed dialogue through this task. The use of this system has shown that the communication paradigm was too narrow to make it a worthwhile tool (especially for the expert user). The SYSTEMS ASSISTANT tries to overcome the noted shortcomings by allowing the users to interact with the system in a mixed-initiative dialogue, to volunteer information and to deviate from the system generated discourse structure.
    TaskMapper BIBA 973-978
      John M. Carroll; Richard E. Herder; Don, III Sawtelle
    This is a description of a prototype office information system called TaskMapper. The key interface idea in TaskMapper is to display the user's activity as a path in a two dimensional map. Thus instead of working with a "messy desk", the user works with a task-oriented layout of recently accessed data. The key system idea in TaskMapper is to represent applications as views on collections of objects drawn from a single database. Instead of working with separate bundles of function and data ("applications") that must communicate constantly to provide even the appearance of integration, the user defines and works with relevant subsets of a single database.

    5. Forefront Systems and Techniques: 5.2 From the Designer's Point of View

    Application Modelling for the Provision of an Adaptive User Interface: A Knowledge Based Approach BIBA 981-987
      E. Adhami; D. P. Browne; S. K. Mitra
    In developing interactive computer systems, the logical separation of the user interface from the application software is a well recognised design principle. The building of adaptive user interfaces requires this separation to be maintained during the specification and implementation stages. Maintaining this separation places special requirements on the communication between the user interface and the application software. This paper discusses the role of application modelling and knowledge based system approach for supporting these requirements. The design and implementation of an application model for constructing an exemplar adaptive user interface are presented and conclusions are drawn regarding the potential benefits of application modelling.
    A Formal Design Methodology for End-User Interfaces -- A Small Case Study Based on UNICON BIBA 989-995
      H. E. Bez; D. J. Cooke
    A unified design methodology [3] currently under study is applied to a small system developed some years ago at L.U.T. [7]. Although simple, the system has been exploited commercially and hence provide a realistic test vehicle for the new methodology. We show how a full formal description of the required interface is evolved by graphical and algebraic means and how it can be 'exercised' by logic programming tools in order to permit the specification to be tested by the system designer. The resulting specification can be used to verify implementations or as starting point for the application of transformational techniques aimed at producing a procedural realisation of the system. The relationship between our approach and other methodologies is also considered.
    Computer Aided Ergonomics Design -- A Program for Suitable Control Locations BIBA 997-999
      Markku Mattila; Markku Leppanen
    Computer Aided Design (CAD) has several advantages over the traditional design process. These advantages may be utilized also when ergonomics and safety are designed. The aim of this paper is i) to describe three CAD-programs with respect to ergonomics in the design of work systems and ii) to present some experiences of the new possibilities offered by CAD to ergonomic and safety improvements in design. The CAD-programs developed and tested in this study are concerning recommended location areas for controls, evaluation of control locations and 3-dimensional man-model. CAD-programs for ergonomic analysis are giving to the designer a new tool to insure the high-quality of ergonomic design.
    Algorithms to Transform the Formal Specification of a User-Computer Interface BIB 1001-1006
      James D. Foley; Won Chul Kim; Christina A. Gibbs
    The Role of the Dialogue System in a User Interface Management System BIBA 1007-1012
      J. L. Alty; J. Mullin
    The respective roles of the Dialogue Controller and Application Model in a User Interface Management System are discussed in the context of a process control application, and application dependent and independent aspects of the dialogue are identified. A multi-channel dialogue controller is proposed which allows concurrent interaction on the interface and dialogues are implemented as a set of dialogue assistants. Application independent assistants control the transfer of information and application dependent assistants implement the task-orientated conversations. As far as possible the design of an assistant is independent of any other. This eases problems in dialogue design and allows assistants to be defined in a specification language which can be analysed for appropriate properties. Assistants can be constructed using a number of paradigms. An event-driven network approach is given as an example. A possible object-orientated approach is briefly outlined.

    5. Forefront Systems and Techniques: 5.3 Advances in Design Techniques

    A Requirements Specification for Next-Generation CAD Systems BIBA 1015-1020
      Phil John
    Current CAD systems suffer from a number of deficiencies. They are typically draughting tools, not design systems, and as such are the logical outcome of bottom-up design, based on assumptions that are no longer sufficient for current manufacturing industry. Instead, by starting from the context in which CAD systems will be used, it is possible to derive a requirements specification that reflects the needs of the organisation. The outcome of this approach is that a great extension is needed to the functionality of CAD systems, and some of the resulting research problems that must be solved before viable systems can be built are outlined. These problems include a discussion of the sorts of organisations in which such CAD systems could be embedded with some chance of efficient use.
    Human Factors in Computer Vision Systems: Design of an Interactive User Interface BIBA 1021-1026
      Volker Haarslev
    A new approach for engineering the user interfaces of image sequence analysis systems is presented. This approach is based on a cognitive model of potential users and on a systematic evaluation of the man-machine communication aspects of image sequence analysis systems. We developed an adaptive user interface for a class of image sequence analysis systems using a data flow architecture. The user interface offers an object-oriented interaction model and allows the direct manipulation of graphical representations of system components. Finally, a corresponding prototype system is described which has been implemented in the programming language Ada.
    The Basis for User-Oriented, Context Sensitive Functions BIB 1027-1032
      James A., Jr. Carter; Michael Schweighardt
    A New Model for Separable Interactive Systems BIBA 1033-1038
      Gilbert Cockton
    Separating the user interface from the 'rest of the application' is a key goal in interactive systems design. The required separation goes beyond software modularity and entails user interfaces and application functions which are both unconstrained by each other's design features. Yet two components which know nothing of each other cannot communicate without an intermediary. A two component interactive system must compromise on separability.
       A new three component model is presented. A linkage component passes user interface information to application functions and returns application function results to the user interface. The second component, the user interface can have a truly generic core. Reconfiguration requires no direct knowledge of application functions. The third component, the non-interactive core, provides the application's functionality. Inter-component relationships are explored to identify user interface features which constrain non-interactive core design and vice-versa.

    5. Forefront Systems and Techniques: 5.4 Advances in Rapid Prototyping

    A Blackboard Architecture for the Realization of Software-Ergonomic Demands BIBA 1041-1046
      Helmut Balzert
    A flexible software architecture is necessary in order to allow the adaptation of the human-computer-interface as well as the application systems to the wishes and abilities of the individual user. The blackboard concept used for the realization of expert systems was modified and extended for the human-computer-interface and the application systems. There is a separate blackboard for the I/O layer, for the dialog layer, for the application systems and for the information & consulting system. The blackboards communicate with each other by tasks. Each blackboard consists of one control part and one domain part. The following knowledge bases have been grouped to the domain part: knowledge about each domain area, user model, user intentions, established conventions, system intentions and self-model. One knowledge source contains local control knowledge, the other contains global control knowledge. The latter controls the strategy to activate one of the existing tasks. The architecture was evaluated in two exemplary implementations.
    Constructive Formal Specifications for Rapid Prototyping BIBA 1047-1052
      Rainer Gimnich; Jurgen Ebert
    The approach presented here suggests a way to translate software specifications into an operational form which can be used as a prototype, for revising the requirements, and for testing purposes by relating it to the actual implementation developed later.
    Rapid Prototyping of Man-Machine Interfaces for Telecommunications Equipment Using Interactive Animated Computer Graphics BIBA 1053-1058
      D. T. Henskes; J. C. Tolmie
    The concept of Rapid Prototyping can be extended from software development support to simulation of man-machine interfaces. This approach will help meeting the challenge imposed on telecommunications engineering by the evolution of an European broadband network system with its subsequent need for highly acceptable user services. Animated computer graphics is a cost effective way for introducing simulation into the earliest possible phase of the design cycle.
    Evaluation of Rapid Prototyping Methodology in a Human Interface BIBA 1059-1063
      J. R. Harris; D. W. Parker
    When designing a custom-built system for a prospective user, they need to understand and relate to the Requirements Specification in sufficient detail to imagine how they will use the system. This has led to the notion of "show don't tell", needing a quick working prototype. Rapid prototyping has been shown to be an essential tool for participative system design.
       We present experiences in developing a prototype for the human interface for a database containing surveillance records for handicapped children, which will be used by clinicians and administrators in a Health Board Authority. The user requirements were: regular assessment of a child's condition; monitoring of the child's access to, and take up of, certain special services needed; and easy compilation of figures relating to the health of the children. We describe the types of changes required by the user in the early phases of the prototype and the limitations which should be imposed on the extent of the rapid prototyping technique. We also discuss the contributions to understanding of the exact needs of the user and how iterative design methodologies achieve those needs. We highlight the benefits that speed and low investment of effort have on the design process.

    5. Forefront Systems and Techniques: 5.5 Advances in Knowledge Based Systems

    Design and Implementation of Direct Manipulative and Deictic User Interfaces to Knowledge Based Systems BIBA 1067-1073
      K. H. Hanne; A. Grable
    The basic ideas for the design and the implementation of a direct (graphical) manipulation interface system are described. Architecture and structure are based on layered models. The system allows the inclusion of deictic / natural language references to objects represented on the screen. The system was developed and implemented on Sun workstations in 'C' and 'Prolog' under the operating system Unix, providing a set of modules and a communication layer for combined interfaces. Two applications, a direct manipulative interface to an expert system and a form query system allowing natural language / direct manipulation interaction, are described.
    X-AiD: An Adaptive and Knowledge-Based Human-Computer Interface BIBA 1075-1080
      Christoph G. Thomas; Gert M. Kellermann; Hans-Werner Hein
    A system will be introduced which realizes an application independent human-computer interface. Its surface offers the users an integrated and extendable structure of icons, windows, menus, natural text, universal operations, and special services. The system X-AiD is designed using AI techniques.
       The interface operates based on declarative knowledge. X-AiD represents common sense about "working with a computer" in general, expert knowledge about the supported applications behind it, and collected knowledge about each of its users. Knowledge is described using the representation language HAL. HAL enables comfortable declaration of object schemes with multiple inheritances and extensive default-handling. All HAL objects may contain specialized rulesets related to human-computer interaction topics (e.g. syntax, semantics, display).
       The system is prepared to work in a "learning by being used"-mode where it memorizes protocols about all occurring actions, including UNDO and REDO operations. Later on it analyses those protocols extracting frequent "plans of action" which the specific user mainly followed. This learned knowledge is applied vice versa to aid this user e.g. by preparing for him situation-dependent menus and objects for him down the mainstream of his work or explaining to him his dialog state and how he got there.
       The working prototype will be presented on a Symbolics 3620 during the conference.
    Metrics for the Building, Evaluation and Comprehension of Self-Regulating Adaptive Systems BIBA 1081-1087
      Dermot P. Browne; Robert Trevellyan; Peter Totterdell; Mike Norman
    Experiences gained during the production of adaptive systems have demonstrated the need for a set of terms by which elements of the system can be referenced. Among these elements are the data being monitored and generated for purposes of user modelling and the control of system behaviour. These data elements, referred to as metrics, are categorised and described in order to provide a terminology that has proved useful during the design, build and evaluation of adaptive systems. The descriptive power afforded by these metrics is discussed by drawing on an example of building a self-regulating adaptive system.
    The Graphical Representation of Knowledge as an Interface to Knowledge Based Systems BIBA 1089-1093
      Ray McAleese
    This paper reports on a technique for portraying the detail and the extent of knowledge on a graphical interface. Research has centered on concept maps which derive from ideas such as "hypertext". Such maps represent the structure and inter-connections between concept labels in knowledge structures. A "map" is a bounded view of one aspect of the overall knowledge-data structure. A map is a synonym for a browser. Users can decompress (expand) or compress (reduce to a minimal level) concept relationships based on selection criteria. This map metaphor is based on cognitive theory that supports the representation of knowledge. In knowledge elicitation the chosen interface gives the expert a network metaphor of their knowledge. Each concept (concept label) can be seen as a label with a finite set of links. As the knowledge structure grows a variety of browsers allow the user to see all or part of the knowledge base. Users can select a topic label as the starting point for a browser, define its limits and specify the type of links (relationships) to be included. Such an interface is a powerful tool as it uses the human eye, which has under used channel capacity, to process complex data structures. The system has been implemented on a workstation which allows multiple windows in a WIMP environment. Knowledge can be entered at any level, compressed (top to down) or decompressed (bottom to up). The system (called KIM: Knowledge and Information Mapping) keeps integrity between different views. The paper reports on user trials with the system. It suggests that problems can arise when the user can not easily reconcile different views and perspectives. The paper highlights the importance of "terrain knowledge" of knowledge based systems (global views), in addition to "street knowledge" (local views).

    5. Forefront Systems and Techniques: 5.6 Novel Application Systems

    Computer Aided Architectural Design Work BIBA 1097-1100
      Lars Kjelldahl; Jerker Lundequist
    This paper deals with the relation between the architect as a professional CAD (Computer Aided Design) system user and the problems of interactive computer graphics. Considerable efforts in R&D are today aimed at looking at how the interaction of professional users with graphic workstations should be designed, to present these users with efficient tools, good working conditions and a choice of ways of using their professional skills.
       The rapid development of computer graphics has created new conditions for design work. The new technology with single user work stations, high resolution screens, efficient tools of interaction, bitmapping, windowing, video input, laser scanning and so on, entails new possibilities for the design professions. Research efforts in the area of man-machine interaction supported by this new technology, involves by necessity several disciplines: computer science, cognitive psychology, linguistics, graphics, esthetics, and design methodology.
       In our everyday language the meaning of the term "design" is vague and complex. Here we use the term "design" as a synonym for "artifact determination". Before the actual production of an artifact can start, it is necessary to determine its properties - of function, form, structure, durability, cost of production and maintenance, as well as to estimate the economic or ecological consequences which will follow. The designer and his client have to determine the properties of the artifact (for example, buildings, consumer goods, machinery, abstract or concrete systems, graphics, that is, all kinds of artifacts or systems of artifacts).
    Research on Model Based Document Processing System DARWIN BIBA 1101-1106
      Miwako Doi; Mika Fukui; Isamu Iwai
    This paper describes a newly developed automatic document architecture extraction system with a document entity model. Hierarchical and anaphoric structural knowledge for chapter and anaphora is represented as a grammar in the document entity model. Automatic document architecture extraction is the first phase of a model based document processing system, DARWIN (Document Architecture Realization for Well-Informed INterface). The purpose of DARWIN is to provide a natural and comfortable document processing environment for both authors and readers. Two other models, an author model and a reader model, are going to be embedded. Experimental results have shown that the automatic document architecture extraction system is sufficiently powerful to detect hierarchical and anaphoric structures. This paper is structured and formatted by the current DARWIN system, in which a simple reader model is implemented.
    How Do We Distinguish the Hyper from the Hype in Non-Linear Text? BIBA 1107-1113
      William P. Jones
    The good news is that non-linear or hypertext systems may dramatically increase the accessibility of information. The bad news is that this increased accessibility may magnify further an already severe problem of selection. Whether we are sending or receiving a body of information, we must take steps to distinguish its components on the basis of their potential importance or relevance. Current hypertext efforts have focused on the development of tools giving users direct control over the formation and traversal of links connecting units of information in a network structure. Such tools place considerable power and a considerable burden in the hands of the users. Information must be initially organized in ways that prove useful later on; links leading to relevant information must subsequently be distinguished from a potentially large number of others. These activities may be very difficult to accomplish in an expanding knowledge base. In this article we look at potential selection in hypertext and we examine some of the ways in which these problems may be remedied.