HCI Bibliography Home | HCI Conferences | INT Archive | Detailed Records | RefWorks | EndNote | Hide Abstracts
INT Tables of Contents: 84879095979901030507-107-209-109-2

Proceedings of IFIP INTERACT'90: Human-Computer Interaction 1990-08-27

Fullname:Proceedings of IFIP INTERACT'90: Human-Computer Interaction
Editors:Dan Diaper; David J. Gilmore; Gilbert Cockton; Brian Shackel
Location:Cambridge, England
Dates:1990-Aug-27 to 1990-Aug-31
Publisher:North-Holland
Standard No:ISBN 0-444-88817-9; hcibib: INT90
Papers:177
Pages:1078
  1. Invited Papers
  2. Foundations: Educational and Social Issues
  3. Foundations: Cognitive Ergonomics
  4. Foundations: User Modelling
  5. Foundations: Formal Methods
  6. Design: Theories, Methods and Tools: Studies and Analyses of Design
  7. Design: Theories, Methods and Tools: Users, Tasks and Organizations: Requirements and Analysis
  8. Design: Theories, Methods and Tools: Prototyping
  9. Design: Theories, Methods and Tools: Evaluation
  10. Design: Theories, Methods and Tools: Design and Evaluation Tools
  11. Detailed Design: Menus
  12. Detailed Design: Graphical and Iconic Interfaces
  13. Detailed Design: User Support
  14. Detailed Design: Hypermedia
  15. Detailed Design: Construction Tools
  16. Interactive Technologies and Techniques: Input
  17. Interactive Technologies and Techniques: Output
  18. Interactive Technologies and Techniques: Speech and Natural Language
  19. Applications and Case Studies: Knowledge-Based Systems
  20. Applications and Case Studies: Computer Supported Co-operative Work
  21. Applications and Case Studies: Applications
  22. Applications and Case Studies: Software Development
  23. Applications and Case Studies: Programming
  24. Doctoral Programme
  25. Panel Sessions
Preface BIB xvii-xviii
  Brian Shackel
Has Done Better -- The Efforts of the '80s in HCI BIBAK xix-xxvi
  Gilbert Cockton
Readers who begin with this introduction and read systematically through all the papers in order will be taken from psychology and mathematics, via design, prototyping, evaluation, interaction techniques and styles, system construction tools and interactive technologies (with real examples all the way) to specific case studies in knowledge-based systems, computer-supported co-operative work, tutoring systems, hypertext journals, aircraft maintenance systems, software engineering and programming. Even though there are uncovered topics and perspectives in these proceedings, the richness of HCI as an area of study and practice should be apparent.
Keywords: Human computer interaction, Multidisciplinary, Interdisciplinary

Invited Papers

SIOIS -- Standard Interfaces or Interface Standards BIBA xxix-xxxiv
  Tom Stewart
National Standards have often acted as barriers to trade, requiring manufacturers not only to produce national variants of products but also to undergo costly and time-consuming certification and testing procedures. After 1992, national standards in Europe will be replaced by European Standards, providing manufacturers with a single European market for their products. In the user interface area, and in many other areas also, European standards do not yet exist and so the European Committee for Standardisation (CEN) plans to adopt International Standards, wherever possible. The purpose of this paper is to provide an update on the activities of the committees working at the European and the International levels to create user interface standards.
Empirical Studies of the Software Design Process BIBA xxxv-xl
  Bill Curtis
Empirical research on software development has shown that the design phase exercises extreme leverage over project outcomes. This paper reviews research performed at MCC on the design process and proposes several research questions whose answers are crucial to improving productivity and quality. One implication of these results is that project outcomes are largely determined before a project begins. (As is the author's habit, the talk at INTERACT'90 may differ substantially from this paper).

Foundations: Educational and Social Issues

Why, What and How? Issues in the Development of an HCI Training Course BIBA 3-7
  Jenny Preece; Laurie S. Keller
Lack of tradition in teaching HCI and the multi-disciplinary nature of the subject creates challenging problems for those developing a curriculum to teach it.
   In this paper we discuss the development of a postgraduate course for practicing professionals in industry and commerce. The course is taught by distance teaching, which means that students study texts integrated with other media on their own.
   Three questions are considered: why we teach HCI, what we teach and how we teach. In addition, we discuss how problems arising from the multi-disciplinary nature of HCI pervaded course development and how the development process itself resembled user-centred HCI design practices.
University Education on Human-Computer Interaction -- The Dutch Situation BIB 9-13
  Gerrit C. van der Veer; Ted N. White
Information Processing, Context and Privacy BIBAK 15-20
  Alan Dix
This paper is about an old concept, data processing, but one that has taken on new meaning with the increasing complexity and interconnection of systems and the burgeoning of expert systems and connectionism. Classical information theory has been found to be inadequate even in the relatively formal context of security, but this inadequacy is intensified when we consider more human issues like privacy. Further, writers like Suchman and Winograd & Flores emphasise context in understanding communication and information. Relating these issues to a simple information life-cycle, this paper questions how we can retain an understanding of human issues when interacting with such complex systems.
Keywords: Information theory, Information processing, Privacy, Formal analysis
Mac-Thusiasm: Social Aspects of Microcomputer Use BIBA 21-26
  Matthew R. Jones
The success of the Apple Macintosh computer is normally ascribed to the quality of its user interface. This paper presents evidence from interviews with Macintosh users of the importance of social and organisational factors in influencing their choice of microcomputer and their pattern of, and attitude to, computer use. Many of the users held strongly positive views about the Macintosh computer and some possible reasons for this enthusiasm are discussed.

Foundations: Cognitive Ergonomics

GOMS Meets the Phone Company: Analytic Modeling Applied to Real-World Problems BIBA 29-34
  Wayne D. Gray; Bonnie E. John; Rory Stuart; Deborah Lawrence; Michael E. Atwood
GOMS analyses were used to interpret some perplexing data from a field evaluation of two telephone operator workstations. The new workstation is ergonomically superior to the old and is preferred by all who have used it. Despite these advantages telephone operators who use the new workstation are not faster than those who use the old but are, in fact, significantly slower. This bewildering result makes sense when seen with the aid of GOMS. With GOMS we can see that very few of the eliminated key-strokes or ergonomic advantages affect tasks that determine the operator's work time. Indeed, GOMS shows that some presumed procedural improvements have the contrary effect of increasing the time an operator spends handling a phone call. We concluded that if GOMS had been done early on, then the task, not the workstation, would have been redesigned.
The Effects of Task Structure and Social Support on Users' Errors and Error Handling BIBA 35-41
  Michael Frese; Felix C. Brodbeck; Dieter Zapf; Jochen Prumper
The relationship of four organizational variables-job complexity, job discretion, social climate, and the organization of the computer advisory service-with number and type of errors and use of support facilities was studied in a field observational study. 198 subjects from 12 different public and private companies in the southern part of the Federal Republic of Germany (secretaries, typists, specialists, low level managers) were observed for 2 hours while they used computers during their work. They also filled out a questionnaire (N=232). There were a number of small but significant and practically important relationships, e.g. errors pertaining to conscious strategies happened more often in more complex jobs. Perceived support by supervisors and co-workers was related to whether these people were asked in case of an error. Decentralized advisory services were preferred and used more often by users than centralized services.
What Kind of Errors Do Unix Users Make? BIBA 43-46
  James H. Bradford; William D. Murray; T. T. Carey
This paper describes a large scale analysis of user error in the Unix environment. Over 300,000 commands were logged containing approximately 16,000 errors. The errors detected included various kinds of token entry problems, mode errors, keyboard errors and grammatical mistakes. The relative frequency of each type is presented. The paper concludes with a discussion of a number of improvements that could be made to the Unix csh command interface.
ECM: A Scheme for Analysing User-System Errors BIBA 47-54
  Paul A. Booth
This paper addresses the question of how best to consider and eliminate the errors that occur during human-computer interaction. Firstly, a distinction is drawn between human errors and user-system errors. A further distinction is then made between different types of user-system error; mapping mismatches and incongruity mismatches. Following this, a classification scheme (an Evaluative Classification of Mismatch, ECM), for analysing user-system errors, is defined, and a study into its usefulness reported. A system that had been refined using ECM was shown to be significantly better, in terms of time, errors and user attitude ratings, than both the original system and a system that had been refined without using ECM. The results of this study suggest that schemes for analysing user-system errors can play a significant role in improving the performance of human-computer systems.
How Learner Characteristics Can Mediate the Effects of Giving Conceptual Details During Training BIBA 55-60
  Elizabeth J. Lloyd
Giving a conceptual model has often been hypothesized to aid learning and subsequent use of computerized devices, but scant attention has been paid to the role of learner characteristics in mediating this training outcome. An experiment was conducted to investigate such aspects further. Two groups of computer-naive subjects were taught to use a text editing system. One group was taught by procedures alone; the other group also received procedures, but in addition was given concurrent conceptual information. The results demonstrated that giving conceptual details during training may not be universally beneficial; overall, there were no great performance differences between the two groups. However, when learner characteristics were examined, it was found that certain individual characteristics interacted with training type to determine the learning outcome. The results imply that conceptual training is only efficacious for certain people, and that it may be possible to predict who those people will be.
Influencing Behaviour Via Device Representation; Decreasing Performance by Increasing Instruction BIBA 61-66
  Simon C. Duff; Philip J. Barnard
Several studies including those of Duff (1989) suggest that prior knowledge plays a crucial role in learning operating procedures and problem solving with computer based systems. This paper tests a counter-intuitive prediction derived from Barnard's (1987) Cognitive Task Analysis. The analysis predicts that the advantageous effects of device knowledge can be offset by providing users with additional procedural instructions during the learning phase. Experimental evidence in support of this prediction comes from users learning to control a simple laboratory application.
Pop-Up Windows and Memory for Text BIBA 67-72
  Heather A. Stark
This study investigates the effect of pop-up windows in hypertext on readers' memory. One group of readers read "plain" descriptions of properties for sale in conventional linear format. Another group read descriptions with the same content, but with some sentences ("new details" about the property) in pop-up windows. Readers judged the suitability of these properties for clients, and then were tested on their memory for property features. For the pop-up group, pop-up property features were recognized more quickly than main text features, and the description they came from was identified more quickly. These sets of features did not differ significantly for the plain group. Pop-ups may be useful ways to emphasize information.
Two Ways to Fill a Bath, With and Without Knowing It BIBA 73-78
  Anne Ankrah; David M. Frohlich; G. Nigel Gilbert
It is generally thought that direct manipulation interfaces are those based on some clear metaphor for interaction which encourages the user to draw an analogy between the machine and some familiar situation. In this paper we challenge this view by showing how it is possible to vary metaphor and directness of manipulation independently in interface design, and that the influence of these factors on usability is not simple. We report findings from an experiment in which 40 users were presented with the same process control task through four different interfaces incorporating different combinations of the two factors. The task was based on the familiar situation of filling a bath.
The Cognitive Dimension of Viscosity: A Sticky Problem for HCI BIBA 79-86
  T. R. G. Green
`Cognitive dimensions' are features of interactive systems considered as information structures, capturing significant commonalities across many types of system. Examples include the presence of asymmetric dependencies; consistency in the command language; and viscosity, or resistance to local change, the subject of this paper. In viscous systems users need many `internal tasks' to accomplish one `external task' of adding, removing, or altering a plan-level component. Opportunistic planning can be thwarted by undue viscosity. Two types are distinguished: Knock-On viscosity, where a goal-related action violates a notational constraint, and requires additional, non-goal-related actions to rectify the problem; and Repetitiousness viscosity, where to achieve a single goal many individual actions are required. Examples of viscosity from many areas are given and some solutions are noted.
The Role of Games and Cognitive Models in the Understanding of Complex Dynamic Systems BIBA 87-92
  Rod Rivers
Computers are being applied to increasingly complex real time monitoring and control tasks. What is the best way to develop this type of computer application? This paper argues that the development of games can reveal a great deal about the complexity of the application, its users and the circumstances in which it will be operated. The development of a game plays a similar role to the development of a prototype or simulation in identifying the most critical factors pertinent to a design. However, developing a game often focuses more attention on the user, his tasks, strategies and learning. Cognitive modelling techniques are especially relevant to understanding the user and contribute to the development of games and to understanding the very nature of complexity.
Implications of Computer Games for System Design BIBA 93-99
  Lisa Neal
Computer games "hold an inexplicable fascination for many people" [Rich 83] and they often provide striking examples of highly motivating activities [Malone 80]. We present a empirical study of game use, focusing on features of games, such as feedback, help, and levels of difficulty, which significantly influence the interaction with a user. We discuss the implications of this research for improving the design of systems, showing that systems not designed for entertainment can benefit from the features incorporated into computer games.
A Study of Measures for Research in Hypertext Navigation BIBA 101-106
  D. G. Hendry; T. T. Carey; S. T. TeWinkel
The research described here investigated the strategies people use to navigate a hypertext document in reading comprehension tasks. We present the results from experiments where people were initially instructed to browse and later to study a document containing expository information on 35 mm. cameras. The users interacted with the document through one of two presentation methods: traditional page sequences or access by hypertext links. A variety of measures were used to illuminate users' navigation strategies: Each of these measures contributes in a different way to our overall understanding of users' navigation (and raises additional questions).
Mental Effort and Task Performance: Towards a Psychophysiology of Human Computer Interaction BIBA 107-112
  David Wastell
Empirical methods play an important role in the science of HCI. The limitations of introspective and behavioural techniques are discussed, using mental workload as an example. The case is argued for a psychophysiological approach to the study of HCI (i.e. the interlocking study of behaviour, experience and bodily processes), adducing as evidence two field studies in which physiological measures played a decisive role. A laboratory study is described in which a physiological measure of mental effort is used to resolve the confounding of effort and cognitive demands inherent in performance measures of workload. The psychophysiological perspective broadens the scope of HCI to include such issues as stress and job design. The paper concludes with a discussion of the possible influence of psychophysiological thinking upon the design of systems for supporting co-operative work.

Foundations: User Modelling

A Knowledge Analysis of Interactivity BIBA 115-120
  Richard M. Young; Andrew Howes; Joyce Whittington
Most existing techniques for predicting users' behaviour do not cope well with highly interactive computer usage, such as is typically found with workstations or personal computers. This paper explores the interactivity inherent in a simple task on an Apple Macintosh computer. An analysis of the knowledge required for performing the task provides the basis for understanding how internal knowledge and information present in the display are combined to guide the behaviour of users spanning the spectrum of expertise from novice to expert.
User's Command Line Reference Behaviour: Locality versus Recency BIBA 121-128
  Alison Lee; Frederick H. Lochovsky
The techniques of working set calculation and locality set determination from computer memory management research are applicable to HCI research. Working sets can be used to analyze the recency characteristic (favouring recent interactions) and locality sets can be used to analyze the locality characteristic (clustering of interactions) in user behaviour. We present two computer simulation studies and their results. The first study found that locality exists in users' command line references. The second study found evidence to suggest that the locality characteristic is better at predicting candidates for recurrence than the recency characteristic. Both of these results have positive implications for the design of tools that allow users to reuse their past interactions.
An Investigation into Quantitative User Modeling of User Interactions for the Purpose of Predicting User Expertise BIBA 129-134
  R. Spall; R. Steele
This paper examines the topic of Quantitative User Modelling. After initial definitions and background work, the potential of computable Quantitative User Models is explored, with particular emphasis given to their application to Intelligent Interfaces. Finally, an example Quantitative User Model is presented in order to highlight the pertinent issues which need to be addressed within this field. This work forms part of an SERC sponsored Research Project being pursued at Sheffield City Polytechnic, which is concerned with the implementation of a complete Object Oriented UIMS.
Constraints in Design: Towards a Methodology of Psychological Analysis Based on AI Formalisms BIBA 135-139
  Francoise Darses
More and more CAD systems are founded on a constraint posting approach. In order to know whether such an approach is compatible with the cognitive activity of the designers, this paper focuses on the crucial role of constraint in the cognitive processes of design. We first highlight how the notion of constraint is referred to in the psychological models of design. A constraint posting approach for design is used to set up a specific methodology of analysis. Based on AI formalisms, this methodology has to be enlarged to give an account of some psychological characteristics of the designers activity. An empirical assessment of the method has been conducted in the domain of computer network design. The results highlight that half of the activity of the experts can be understood in terms of constraint posting. Moreover, we show that the ability to handle the constraints differs, depending on the level of expertise of the designers and according to the nature of the constraints. But above all, this assessment lays down the limits and the benefits of the methodology. It sketches the directions in which the method will have to be enlarged.

Foundations: Formal Methods

Agents: Communicating Interactive Processes BIBAK 143-148
  Gregory D. Abowd
In this paper we extend the theory of communicating interactive processes based on the formalisms of CSP and Z first presented by Sufrin and He. We then use this theory to show how specifications of complex interactive systems are composed from smaller and simpler components. The theory provides a formal foundation for the investigation of multiagent models and allows the precise formulation of design principles for the development of interactive systems. We view an interactive system as a closed collection of communicating interactive processes, or agents. An agent can represent both an application and its users. We develop an example of a mouse and keyboard input to a simple display manager. This will serve as a small but realistic example to explain the features of the model.
Keywords: Interaction model, Formal methods, Software architecture
Pattern Recognition and Interaction Models BIBA 149-154
  Janet Finlay; Michael Harrison
Human Computer Interaction can usefully be described in terms of a sequence of user and system events. A priori traces of such event sequences, as specified by a mathematical model, can be used in the evaluation of interactive systems by contrasting them to a posteriori traces of actual user behaviour. We use pattern recognition techniques to automate this comparison, identifying points in the interaction where a user's behaviour is sub-optimal. We describe work in this area relating to a bibliographic database system.
Formal Analysis of Co-Operative Problem Solving Dialogues: Tools and Techniques BIBA 155-161
  P. Jeremaes
We are developing task specific theories of co-operative problem solving and using these theories to derive methods for engineering knowledge-based systems which support co-operative human-computer interaction. Research to date has concentrated on the collection of empirical data to support the hypothesis that co-operativity in this context is warranted and to develop our initial theories of co-operative interaction. This paper presents an overview of the role that formal tools and techniques have played in helping to analyse that data and in formulating tentative theories of co-operation.
ETAG: Extended Task Action Grammar -- A Language for the Description of the User's Task Language BIB 163-168
  Michael J. Tauber
ETAG -- Some Applications of a Formal Representation of the User Interface BIBA 169-174
  Gerrit C. van der Veer; Diederik Broos; Kenneth Donau; Mark J. Fokke; Felix Yap
Based on the theoretical work described in the contribution by M.J. Tauber (this Volume), we conducted several studies aiming at the application of the ETAG formalism. We concentrated on 3 aspects of user interface design and analysis:
  • The application of ETAG based evaluation methods for the analysis of existing
       systems from the point of view of learnability and usability;
  • Formal representation methods like ETAG based formalisms, for the analysis
       and formal description of user's tasks;
  • The use of ETAG and related formalisms for the design of user interface
       modules (automatic generation of on-line manual and help text, and modules
       for answering users' questions about the system).
  • What is Inconsistency? BIBA 175-181
      Phyllis Reisner
    One of the basic tenets of interface design is that an interface should be consistent. However, the meaning of the term remains elusive. Recently, there have been several attempts to represent inconsistency formally. A goal of such representation is to identify the inconsistency automatically. This paper (1) shows how each new formalism has increased our understanding of the nature of inconsistency, by expressing assumptions that had been hidden in earlier ones, (2) argues that none of the current formalisms will be able to identify consistency reliably because of an assumption that is still missing, and (3) presents a formal framework, APT, which includes the missing assumption. APT (Agent Partitioning Theory) is used to explain why user errors - a result of inconsistency - occur.

    Design: Theories, Methods and Tools: Studies and Analyses of Design

    Looking HCI in the I BIBA 185-191
      Stephen J. Payne
    Despite its name, the field of human-computer interaction has not devoted much research attention to the nature of interaction. This paper begins such an effort by focussing on the role of the user interface as a resource for action. A simple notation for describing human-computer interactions, based on Clark and Schaefer's (1987, 1989) theory of conversational contributions is motivated and described. Interaction trees allow descriptions that expose the role of the device's detailed output dynamics in user activity.
    Qualitative Artifact Analysis BIBA 193-198
      Wendy A. Kellogg
    The psychology of HCI in the 1990s will be concerned with embodied cognition and action: helping us understand real people acting in real situations, and doing so in a way that enhances existing design practice. One proposal for meeting these objectives is qualitative artifact analysis: specifically, creating psychological design rationales by extracting claims from scenarios of use. I describe our method for doing this and its key requirements, and suggest how artifact analyses of this kind can serve to cumulate knowledge about usability in a form useful for design. A critical requirement for deriving benefit from analyses of human-machine interaction is expressing understandings and usability outcomes in terms of the artifact's design. In support of this, I reconsider an example of the analysis of situated action from the point of view of artifact analysis.
    Redesign by Design BIBA 199-205
      Rachel K. E. Bellamy; John M. Carroll
    The evolution of HCI technology has been characterized by the task-artifact cycle, raising the question how do artifacts change tasks? In answer we have started to analyze personal redesigns of Smalltalk/V tools, to understand how and why the programming environment and the tasks it supports evolve. We interviewed designers working with Smalltalk/V, and asked them to describe their personal redesigns of the system tools (browsers, inspectors etc.), motivations for redesign, and typical scenarios of use before and after redesign. We found that designers do consider usability issues, but sometimes not all the usability strengths and weaknesses of existing interface techniques and of their redesigns. We show how psychological claims analysis can support and guide such redesign work.
    What Rationale is There in Design? BIBA 207-212
      Allan MacLean; Victoria Bellotti; Richard Young
    Design Rationale is a framework for locating a proposed design within a design space. It incorporates an explicit representation of design Options, and an explicit representation of Criteria for choosing among the Options. This paper explores the relationship between Design Rationale and design practice. It uses Design Rationale as a way of analysing the content of a design session to help us understand requirements for future ways of improving the design process.
    A Framework for Assessing Applicability of HCI Techniques BIBA 213-218
      Victoria M. E. Bellotti
    The findings from three studies of applied and commercial design practice provide the basis for a framework for assessing the applicability of HCI analytic techniques. This framework embodies an explicit view of the design process, HCI oriented design roles, and a scoping matrix designed to represent breadth of a design or evaluative approach. These components assist in the identification of a list of desirable features for more applicable techniques, derived from interviews with practising HCI specialists in commercial software houses.
    Obstacles to User Involvement in Interface Design in Large Product Development Organizations BIBA 219-224
      Jonathan Grudin
    Development of an "off-the-shelf" product typically starts with a product idea and limited knowledge of the eventual users. Since the functionality is partially predefined, the most natural focus for user involvement in design is the human-computer interface. However, large product development organizations contain inherent obstacles to involving existing or potential users even in interface design. Formed before the human-computer interface attained its present prominence, their organizational structures and development processes have evolved with minimal consideration of the particular needs of interface development. This paper outlines the problems in achieving and benefiting from user involvement in design that stem from typical divisions of responsibility and development processes. While overcoming such organizational constraints may ultimately require organizational change, those working within such an organization must be aware of the problems and constantly seek constructive paths around them.
    Integrating Human Factors with Structured Analysis and Design Methods: An Enhanced Conception of the Extended Jackson System Development Method BIBA 225-230
      K. Y. Lim; J. B. Long; N. Silcock
    The potential benefits of integrating Human Factors (HF) with structured analysis and design methods have been described previously, along with an initial conception for integration instantiated using the Jackson System Development (JSD) method (termed JSD*). Based on a case-study test, the initial conception has been enhanced by further extensions and developments of its HF design stream (termed JSD*(HF)) and these are reported in this paper. The case-study which involves the re-design of a recreational facility booking system (RFBS) is used to illustrate the enhanced conception. It is concluded that the conception shows promise. Future developments of JSD*(HF) are also proposed.

    Design: Theories, Methods and Tools: Users, Tasks and Organizations: Requirements and Analysis

    An Investigation of User Requirements for Broadband Communications in the Automotive Industry BIBA 233-238
      S. E. Powrie; C. E. Siemieniuch
    This paper discusses a user-centred research methodology for the investigation of the role for multimedia integrated broadband communications (IBC) in the automobile industry. The derivation of a comprehensive user requirements set is described, key requirements outlined and the usefulness of the methodology itself considered.
    Bridging the Gap Between Task Design and Interface Design BIBA 239-245
      W. Dzida; R. Freitag; C. Hoffmann; W. Valder
    "Work context" is an ergonomic and technical concept for user interface development. Essentially, interface development is taken to mean more than screen layout, explanatory dialogue or something like that; it is the user's way of performing a task which is developed. Software developers are the particular user target group. For the technical and ergonomic ideas to become illustrated, the prototype of a user interface -- called ERGO-Shell -- has been constructed. The shell assists the user to apply a plan-oriented style of dialogue in that the system provides facilities for work preparations in complex task settings. Thus, ERGO-Shell enables the user to interact with the system on a demanding level.
    Supporting a Humanly Impossible Task: The Clinical Human Computer Environment BIBA 247-252
      B. Horan; A. L. Rector; E. L. Sneath; C. A. Goble; T. J. Howkins; S. Kay; W. A. Nowlan; A. Wilson
    Medicine has proved a fruitful field for developing knowledge based systems. Paradoxically, the General Practice medical environment has a number of characteristics which make the introduction of such systems difficult. Attempts to produce systems for other professional users -- e.g. architects, lawyers, and executives -- have had somewhat similar experiences. However, doctors work under severe time pressure in a complex social environment. The neatly confined problems most tractable to expert systems have limited relevance to doctors' decision making in practical situations. Furthermore, doctors already have a well developed system of sharing expertise.
       Extensive user centred design studies have led us to propose an alternative model for augmenting doctors' performance. Rather than an expert system, we propose an intelligent human-computer environment for maintaining medical records and `throwing light' on the complex data of patient histories.
    An Analysis of the Circuit Design Process for a Complex Engineering Application BIBA 253-258
      Lynne Colgan; Maddy Brouwer-Janse
    An approach for improving human-computer interaction in a complex engineering domain is presented. A multi-disciplinary team addresses the issue of finding an appropriate environment for analogue circuit optimisation. This environment has to bridge the gap between users in the electrical engineering field and numerical techniques. Our solution is to adopt a long term system-design approach integrating cycles of prototyping with user evaluations. This approach highlights the conflict between designing a controlled psychological experiment from which conclusions can be drawn with statistical confidence, and a practical, incremental system development. The general philosophy of the approach is to enlist users tackling realistic problems in everyday settings.
    Designers-Identified Requirements for Tools to Support Task Analyses BIBA 259-264
      Hilary Johnson; Peter Johnson
    The authors are presently developing tools to enable software designers to carry out task analyses (TA). The tools will support a methodology comprising techniques for carrying out task analyses and will also take account of integrating the resulting TA information into system design. To support integration and to identify the requirements for TA tools, a group of designers were surveyed. The survey identified whether designers believe TA would be of use to them and also how, why and where TA might contribute to design. The designers' views of desired characteristics of TA tools, was also sought. This paper outlines the results of this small, detailed survey of what designers want, need and expect from TA tools.
    An Application of Task Analysis to the Development of a Generic Office Reference Model BIBA 265-269
      Jill Hewitt; John Hobson; John Sapsford-Francis
    The roles of task analysis in the software engineering life cycle are considered, and a method is described which is suitable for capturing the high level communication tasks in offices. The contribution of this method to the building of a generic office model and its role in the generation of scenarios for future early requirements analysis are discussed.
    Memory-Cognition-Action Tables: A Pragmatic Approach to Analytical Modelling BIBA 271-275
      Brian Sharratt
    Current analytical models have a number of practical problems -- they use complex notations, produce very detailed task descriptions and tend to be difficult to use for anything other than small examples. This paper describes a simple analytical model based on multi-columned tables, called Memory-Cognition-Action (MCA) tables which is used to address these problems. The paper concentrates on the production of MCA tables and the analytic metrics derived from these tables. The metrics cover task complexity, cognitive processing and task consistency and can be used to locate potential user problems in the task structure. Features of the MCA tables such as their extendability, reusability and handling of task closure are discussed.
    Analysing Focused Interview Data with Task Analysis for Knowledge Description (TAKD) BIBAK 277-282
      Dan Diaper
    While Task Analysis for Knowledge Descriptions (TAKD) is now a reasonably well established task analysis method, its use to date, in the published literature, has been principally concerned with the analysis of observational data from task performance exercises. The paper describes the use, in an industrial context, of TAKD to organise data from interviews, albeit where there was a focus on the interviewees' tasks and subtasks.
    Keywords: Task analysis, Interview data, Automatic test equipment
    A Plan & Goal Based Method for Computer-Human System Design BIBA 283-288
      Daniel R. Sewell
    A methodology for designing computer-human systems based on explicitly representing the user's hierarchical plan and goal structure is presented. The methodology provides an advance over traditional task-analytic methods by providing representational structure that captures both causal relationships of the operational domain and cognitive states of the system user. The plan-goal graph representation maps directly to system functions and actions, aiding both design and implementation of the system. This methodology is illustrated with a review of a recent application to designing an information retrieval and analysis system for system designers attempting to access and apply human performance information to design problems.
    The Use of Task Allocation Charts in System Design: A Critical Appraisal BIBA 289-294
      W. K. Ip; L. Damodaran; C. W. Olphert; M. C. Maguire
    The design of IT systems has traditionally failed to take account of the job design requirements of end users, resulting in negative and unplanned effects on user acceptability and system efficiency. A technique based on the idea of Task Allocation Charts was developed to enable job design issues to be represented and considered in the early stages of IT system design. This attempts to address the problems of Requirements Capture and the limitations of traditional Systems Analysis in identifying job requirements in system specification. A description of the notation and usage of the charts is followed by a critical appraisal of their application in a real design situation. The paper concludes with proposals of future research and the development of a CASE tool to support the technique.
    The Development of Tools to Assist in Organisational Requirements Definition for Information Technology Systems BIBA 295-300
      S. D. P. Harker; C. W. Olphert; K. D. Eason
    The 5-year CEC ESPRIT II Project 2301, begun in January 1989, aims to create a methodology called ORDIT (Organisational Requirements Definition for Information Technology systems) which will enable those engaged in systems analysis and design to identify and address organisational requirements. The methodology will use concepts at the organisational and work role levels of description to represent these requirements, and will also provide a variety of tools (including software tools) for using this representation as a simulator to explore the implications of systems at various stages in their development. This paper discusses the concepts which underpin the Project, the emerging methodology, and prototype tools.

    Design: Theories, Methods and Tools: Prototyping

    Hypermedia as Communication and Prototyping Tools in the Concurrent Design of Commercial Airplane Products BIBA 303-308
      Elfriede Hofer; Frank Ruggiero
    This paper discusses the utility of hypermedia technology as a communications and prototyping tool for a "concurrent" or "integrated" product design approach. This design approach uses multifunction teams, called Design-Build-Support Teams (DBST's), with members from the design, manufacturing, and customer service disciplines working together from the outset to communicate effectively and produce a robust design -- one which requires a minimum of change throughout the product's full life cycle. Examples of hypermedia-based simulation applications from the commercial airplane domain are discussed. These applications include a B747-400 full-flight simulator instructor control station, an animated simulation of the B737-300 airplane pneumatic system and a user interface prototype of an avionic module.
    An Object-Oriented Framework for Prototyping User Interfaces BIBA 309-314
      Peter Windsor
    We have developed a software framework for prototyping user interfaces combining the technology of user interface management systems and window system toolkits. The architecture provides: support for the overall structure of the interface, the means to define and control the dialogue at high and low levels and control over the fine details of presentation and interaction style. The design is object-oriented and achieves great flexibility through inheritance and polymorphism. The framework has been used to produce operational prototypes as part of the requirements analysis for a new workstation for the Oceanic Air Traffic Control Centre.
    Paper versus Computer Implementations as Mockup Scenarios for Heuristic Evaluation BIBA 315-320
      Jakob Nielsen
    A taxonomy of the various forms of scenarios in the user interface field is given, including a discussion of different forms of mockups. A single interface design for a videotex system was implemented as a mockup in two different mediums: As a paper mockup and as a running prototype using HyperCard. These two versions of the same design were then subjected to heuristic evaluation by two similar groups of evaluators. Both versions contained the same fifty usability problems, but there were great differences in the types of problems found by the two groups of evaluators. This indicates that the medium in which a design is presented will have a major impact on what kind of usability problems can be discovered using heuristic evaluation.

    Design: Theories, Methods and Tools: Evaluation

    Evaluating the Usability of User Interfaces: Research in Practice BIBA 323-328
      Arja Vainio-Larsson
    A prototype for a network operation and control system has been evaluated as part of a study on methods for evaluating usability. Interviews, direct observations, video recordings and thinking aloud techniques has been employed to collect information from users. Additionally a checklist was used as a guide for an expert assessment of the prototype. The methods were chosen on the basis of suitability for data collection in a field study. The approach generated considerable amounts of data, and several usability problems were identified. However, analysis and compilation of the data was time consuming, difficulties were also encountered in feeding back the results of the evaluation into the design process. Since evaluation is more than merely an exercise in data collection these issues must also be considered in selection of evaluation methods.
    Evaluating Evaluation: A Case Study of the Use of Novel and Conventional Evaluation Techniques in a Small Company BIBA 329-335
      Jonathan Crellin; Thomas Horn; Jenny Preece
    During recent co-operative working with an industrial partner, a number of usability evaluation techniques were compared in an evaluation of seven interface prototypes. A detailed real-time software log of the interaction was recorded and video and audio records of the interactions were kept. Additionally, a novel experience elicitation technique, based on knowledge elicitation techniques was used. Keystroke and BNF analyses were also prepared. This material has allowed a comparison of the different methods to be made, and recommendations as to their use to be drawn up.
    Integrated Office Software Benchmarks: A Case Study BIBA 337-343
      James R. Lewis; Suzanne C. Henry; Robert L. Mack
    In this paper we present a case study of a benchmark evaluation of integrated office systems. The case study includes developing scenarios, benchmark measures, and quantitative and qualitative analysis of user performance and user problems. We studied two systems, one loosely integrated windowing environment and one more tightly integrated (with respect to consistent graphical interface style). Multivariate analyses showed that significant differences were attributable to performance/analytical variables and to patterns of error impact classifications, but not to subjective ratings. Somewhat surprisingly, users experienced serious problems with the seemingly more integrated (consistent) system largely because of a handful of serious problems. This was taken as evidence that improvement of the poorer performing system should be based primarily on an analysis of errors. Some examples are presented to indicate the potential diagnostic value of analyzing of problems and the development of testable behavioral objectives from benchmark measures.
    Comparative Study of Geometry Specification Capabilities of Geometric Modelling Systems BIBA 345-350
      K. Case; B. S. Acar
    This paper describes experimentation carried out with a novel computer aided design system which uses 'manufacturing features' as the principal method by which the designer specifies the geometric part of design. The approach has been experimentally compared with the more traditional methods of two dimensional computer aided draughting and solids modeling. With skilled industrially-based design engineers such a features approach compares well with the other methods in terms of the user time and the accuracy and completeness of the final models.
       Learning studies in the new technique and a large scale attitude survey are also briefly described.
    Cost-Benefit Analysis of Iterative Usability Testing BIBA 351-356
      Clare-Marie Karat
    A methodology for computing the value of iterative usability work is presented using data from a series of three usability tests of a software application. The cost-benefit analysis methodology provides software development managers a basis for making pragmatic decisions about human factors work. The projected dollar value of the reduction in end user time on an application task based on data from the first to the third test is compared to the costs of the usability work. The analysis of initial end user application use shows a two dollar return on every dollar invested in usability project activities and highlights sources of additional savings. Two methodological techniques employed during the iterative usability testing are highlighted and the decision process concerning use of these techniques for human factors, software development schedule, and economic reasons is discussed.
    Usability Statements and Standardisation -- Work in Progress in ISO BIBA 357-361
      John Brooke; Nigel Bevan; Fred Brigham; Susan Harker; David Youmans
    This paper describes work in progress in Working Group 5 of the International Organisation for Standardisation Technical committee 159 subcommittee 4 (ISO TC159/SC4/WG5). While many standards are concentrating on what guidelines can be given regarding the design of user interfaces and dialogues, subgroup 2 of TC159/SC4/WG5 is taking a holistic approach to the issue of the usability of products. A standard is being developed which will specify how producers and consumers of products may communicate with each other about the usability of products.

    Design: Theories, Methods and Tools: Design and Evaluation Tools

    Supporting Effective and Efficient Design Meetings BIBA 365-370
      John Karat; John Bennett
    We present a description of the methodology employed in our evolving user-centered design framework. Our experience suggests that the quality of human-computer-interaction (HCI) supported by a system design will be strongly influenced by the insight generated in design work at early stages of development. Our methodology focuses on creating a shared vision and working environment within the design team for productive action on system objectives, constraints, resources, and proposed designs. Two aspects of the methodology which we focus on are use of the walls of a "design room" to hold representation of the design-in-progress, and fostering collaboration through discussions by team members with different skills and perspectives. Our experience indicates that the framework and techniques used within it are generally applicable, but that successful use requires attention to characteristics of each particular design project and group.
    The HUFIT Planning Analysis and Specification Toolset BIBA 371-376
      Bronwen Taylor
    The Planning Analysis and Specification (PAS) Toolset has been developed to improve the requirements capture and specification of IT products. The tools are structured techniques which help to develop User and Task information into appropriate product requirements specifications. They have been developed iteratively, in collaboration with product Planners, Marketeers, Developers and Testers in large European IT manufacturers for use by these groups in the planning of generic IT products. The tools present a user-centred approach to design with a number of beneficial features: they improve communication in design teams, structure and support design decisions about user needs and record information for reference by other participants in the design process. Experience in training designers and human factors practitioners in the use of the tools has confirmed the usefulness of this approach.
    The HUFIT Functionality Matrix BIBA 377-381
      Bernard J. Catterall
    The Functionality Matrix, part of the HUFIT Planning Analysis and Specification Toolset for IT Product Designers, cross references user requirements information with the technical proposal in a first-pass assessment procedure which seeks to arrive at an enhanced functional specification. In business terms, the Matrix provides an explicit mechanism for assessing user and task-related issues within the context of the other business trade-offs which are made during real product development.
    Task-Based User Interface Development Tools BIBAK 383-387
      Peter Johnson; Emma Nicolosi
    The generation of design ideas can be facilitated by user/task analysis. Task analysis can influence the design of functionality, dialogue and presentation characteristics of user interfaces. A case history of designing a user interface to a CAD system using Knowledge Analysis of Tasks (KAT) is reported. Methods and tools to assist designers in carrying out user/task analysis have been developed and are described. These include a hypercard task simulation tool (DETAIL) and a task based prototyping tool which allows user interfaces to be developed from task analysis data.
    Keywords: Task analysis, Prototyping and simulation tools, Computer aided design
    PROTEUS: An Approach to Interface Evaluation BIBA 389-394
      Jonathan Crellin
    PROTEUS is a number of software tools which allow the implementation of an iterative, user centred approach to software (particularly interface) design, using rapid prototyping. The tools allow automated collection of questionnaire data, logging of system usage, and the central technique which is the collection of a qualitative representation of users perception of an interface design space, using the Construct Elicitation System. This data is fed back to the designer, and increases understanding user needs in relation to an interface. The paper describes the development of PROTEUS as an integrated evaluation tool, and reports on some of the empirical work underlying the approach embodied by PROTEUS, including its integration into the design of a small but real system.
    A Knowledge-Based Tool for User Interface Evaluation and its Integration in a UIMS BIBA 395-400
      Jonas Lowgren; Tommy Nordqvist
    This paper describes and discusses a knowledge-based user interface evaluation tool, based on the critiquing paradigm. The tool uses knowledge acquired from experts and from collections of guidelines to evaluate a formal description of a user interface design, generating comments as well as suggesting improvements.
       After describing the system architecture and reporting some experiences, the paper focuses on the possibility of incorporating a knowledge-based design tool in a User Interface Management System (UIMS), making it possible to give constructive advice to the designer as well as comments. We report some preliminary results from a project aimed at this integration.
    Monitoring and Analysis of Hypermedia Navigation BIBA 401-406
      Diana Kornbrot; Miles Macleod
    The use of an interaction monitoring tool in conjunction with commercial spreadsheet and statistical packages is described. The tool was used to monitor and analyse M.Sc. students' use of a hypermedia system with multiple navigation structures to study course content. The final product of the analysis is a description of the navigation routes and methods used by individual students to acquire information from the courseware. Post hoc, students were clearly separable into those who performed relatively more, and those who performed relatively less, actions per minute. These two groups were also different in terms of their use of the available navigation structures and the content they chose to visit. The role of high level monitoring tools and associated analysis packages in evaluating hypermedia material, and in answering questions about human learning, is discussed.
    Towards an Evaluation Planning Aid: A Feasibility Study in Modelling Evaluation Practice Using a Blackboard Framework BIBA 407-413
      Ian Denley; John Long
    This paper assesses the feasibility of the blackboard architecture as an organisational schema with which to model evaluation practice, and to provide an initial input to the development of an evaluation planning aid for practitioners. The paper illustrates the potential of a blackboard framework as a structure for making explicit the classes of knowledge used by human factors practitioners in the evaluation of interactive human-computer systems. A number of case histories of evaluation practice are modeled in terms of the framework, and provide examples of its applicability. It is concluded that the blackboard architecture has potential as a structure with which to model evaluation practice.
    Providing Intrinsic Support for User Interface Monitoring BIBA 415-420
      Jolly Chen
    Effective user interface evaluation requires unobtrusive recording of dialogue data. This paper examines how dialogue information can be acquired from monitoring the communication channels between different linguistic levels in the user interface model. Adding built-in monitoring mechanisms to the architecture permits recording of dialogue information not otherwise accessible from low level input recording techniques. Intrinsic support also allows an application program to be monitored without modifying the application source code. Enhancements have been made to the Xt Intrinsics architecture as a sample implementation of this approach.

    Detailed Design: Menus

    Are All Menus the Same? -- An Empirical Study BIBA 423-427
      Zsuzsanna Mills; Martin Prime
    In the "direct manipulation" style of interacting with computers pop-up menus are becoming increasingly popular. The present study looks at the speed and accuracy of six different menu styles falling into two main groups, "moving" and "static" menus. After an initial "ballistic response task", subjects carried out a block of 15 selections with each menu style. The order of the menu styles was randomly varied for each subject to counter-balance possible "fatigue" and practice effects over subjects. The order of the items on the menus was also randomised to force subjects to use visual search at each trial rather than relying on their memory of the item's position from previous trials. The analysis of the response latencies for menu item selection indicates a clear performance advantage with static menus. There was no interaction between "skill level", determined by the ballistic response task, and performance on the menu selection task. The fastest and least error prone amongst the menu styles proved to be the circular menu.
    Pull-Down, HoldDown, or StayDown? A Theoretical and Empirical Comparison of Three Menu Designs BIBA 429-434
      Miles Macleod; Penelope Tillson
    Pull-down menus can be cumbersome to use when making multiple choices, as they become hidden after each choice. They may also be criticized for paucity of feedback about choices made. This paper considers two alternative designs, which help overcome these shortcomings: a menu which can be set to stay visible until closed by the user; and a menu which can be held in view while required, by pressing a 'hold' key. The user actions required by these design alternatives are evaluated theoretically, with the help of user action notations, and predictions generated about some aspects of usability. The implementation in HyperCard of working, self-monitoring prototypes is described. An empirical comparison for usability of the implemented designs is reported, where the StayDown and HoldDown menus were found to be significantly faster than a pull-down menu for making multiple choices, and to be subjectively preferred, especially for their enhanced feedback about currently chosen attributes.
    The Use of Guidelines in Menu Interface Design: Evaluation of a Draft Standard BIBA 435-440
      Flavio de Souza; Nigel Bevan
    This paper reports a case study of the evaluation of the effectiveness of a draft standard containing human factors guidelines for menu interface design. Three designers were given a week to study the guidelines, before spending a day using the guidelines to redesign a menu interface. They were asked to justify their design in terms of the guidelines. The designers made errors or had difficulties with 91% of the guidelines. The cause of the errors and difficulties was analysed, which enabled recommendations for improvements to be made. Despite the difficulties with interpretation of the guidelines, the resulting interfaces only violated an average of 11% of the guidelines which could be assessed. It is concluded that it was difficult for the designers to integrate detailed design guidelines with their existing experience.
    Decision Track: A Formalism for Menu Structure and User's Selection Behaviour BIBA 441-446
      William Edmondson
    The paper presents a new notational formalism designed to permit the recording of sequences of selections, and also to express the space of possible selections, in menu-based interfaces. The fundamental insight is that when people make selections from a menu they do so as part of a process of decision making; selections are organized into sequences, and it is these which characterize the expression of decisions. It is intended that the notation will be used as the output of task analysis and as the input to both the formalism of Lean Cuisine (Apperley & Spence, 1989), and the formalism of User Action Notation (Siochi & Hartson, 1989). Decision Track is one of the formalisms in the proposed framework for Systematic Menu Design (Edmondson and Spence, 1990).

    Detailed Design: Graphical and Iconic Interfaces

    A Three-State Model of Graphical Input BIBA 449-456
      William A. S. Buxton
    A model to help characterize graphical input is presented. It is a refinement of a model first introduced by Buxton, Hill and Rowley (1985). The importance of the model is that it can characterize both many of the demands of interactive transactions, and many of the capabilities of input transducers. Hence, it provides a simple and usable means to aid finding a match between the two.
       After an introduction, an overview of approaches to categorizing input is presented. The model is then described and discussed in terms of a number of different input technologies and techniques.
    Iconic Interfacing: The Role of Icon Distinctiveness and Fixed or Variable Screen Locations BIBA 457-462
      Alison J. K. Green; Philip J. Barnard
    This study examined the ease with which icons differing in visual distinctiveness are learned and searched in either fixed or variable screen locations. Previous research by Arend, Muthig and Wandmacher [4] found that with random arrays, abstract icons were searched faster than representational icons. The present experiment manipulated the degree of locational ambiguity within arrays of abstract and representational icons in order to identify general principles governing the learning and searching of icon arrays. Results clearly show that differences between search times for abstract and representational icons are substantially reduced with arrays in which the position of all icons remained fixed. These and more detailed findings are used to frame constraints which may be governing cognitive activity in search and select tasks.
    Where to Draw the Line with Text: Some Claims by Logic Designers about Graphics in Notation BIBA 463-468
      M. Petre; T. R. G. Green
    Is graphical notation really superior to text, or just different? This paper reports observations of professional hardware designers and considers claims they make about graphics and text on a variety of issues: overviews, zooming, adjacency, detail, viscosity, searching, and space consumption. This paper concludes that the key factor in choosing between graphics and text is the accessibility of information demanded by the user's task.
    The Power of Parameterizable Objects in Modern User Interfaces BIBA 469-472
      Franz Penz; Manfred Tscheligi; Gunter Haring; Martina Manhartsberger
    Design alternatives for modern object based user interfaces are presented trying to reach a second generation of enjoyable graphical user interfaces. The objects are designed following the real life paradigm, yielding to intuitive human computer associations. Objects are parameterized to give the user some possibilities to adjust his operating environment, according to his special needs. Menus, highly used in traditional user interfaces, are superfluous in our introduced system.
    Alternative Bases for Comprehensibility and Competition for Expression in an Icon Generation Tool BIBA 473-477
      Stephen W. Draper; Kevin W. Waite; Philip D. Gray
    We have constructed an icon generation tool (called "Iconographer"), as reported in a companion paper [1]. It can be regarded as possessing multiple inheritance of research areas: either as a user interface management system (UIMS) narrowly specialised on iconic presentation, or as a data visualisation package specialised to apply to user interfaces. There are similarly contrasts in whether the icons are hand drawn or generated, and in the sources of knowledge which users bring to bear in interpreting graphics: resemblance to known objects (e.g. little drawings of printers), or comparison with the other icons in view and with axes and keys drawn on the background. The work of Gaver and J. J. Gibson suggests the possibility of a closer synthesis of these approaches.
       The information competing for expression comes from several distinct general sources. We distinguish three: the central arena of system objects offering system properties such as file size, the semantic arena of concepts coded by the user such as "all files to do with my current project", and the articulatory arena concerned with interactive properties such as "this object may be clicked on".
    Integrating Natural Language and Graphics in Dialogue BIBA 479-484
      John Lee; Henk Zeevat
    Natural interfaces to "intelligent" systems have much to gain from the integration of natural language and graphics. Neither medium is sufficient alone. Experience in the development of an integrated interface (the ACORD demonstrator system) indicates that graphical visualisation must include syntactic specifications of interactions and their semantic interpretations. These issues are discussed in the context of moves towards unifying the specifications of graphical interactions and natural language sentences in a single formalism.
    Semantics and Graphical Information BIB 485-491
      Ewan Klein; Luis A. Pineda
    Using Depictive Queries to Search Pictorial Databases BIBA 493-498
      Stephen Charles; Stephen Scrivener
    This paper argues that pictorial databases are becoming, and will continue to be, important in information systems because pictures can be used to depict information which is difficult to describe or perhaps incomprehensible in non-pictorial form. It follows that when searching for a picture the user might find it easier to depict the query by means of a picture. This paper describes a method for searching pictorial databases where the user essentially constructs a sketch (which combines depiction and description) of the target picture.
    HyperBliss: A Blissymbolics Communication Enhancement Interface and Teaching Aid Based on a Cognitive-Semantographic Technique with Adaptive-Predictive Capability BIBAK 499-503
      Ami Shalit; David A. Boonzaier
    Blissymbolics is a semantically-based graphic language. It is internationally accepted as a comprehensive and effective alternative communication system for severely disabled people. A new approach to the selection of Blissymbols has been implemented as a HyperCard application for the Apple Macintosh computer. The result is a user interface designed to enhance communication through Blissymbolics and facilitate the learning thereof. Following an introduction to the context within which HyperBliss has been developed, some of the features which are unique to this programme are illustrated and discussed.
    Keywords: Blissymbolics, User interface design, Augmentative and alternative communication, Semantics, Graphic language, Special education
    A Cognitive Approach to the Definition and Evaluation of a Standard for Naval Tactical Display Symbology BIBA 505-512
      John Campion; Martin A. Brockett; Dan Martin; Michele Rate
    A cognitive approach to systematically and objectively defining and evaluating options for a standardised coding scheme for naval tactical displays was described and illustrated through the partial development of a method. The method comprises three parts; a consideration of the process of defining the standard, an evaluation framework for the coding schemes which considers system effectiveness and the general nature of command level tasks associated with tactical displays, and a task model derived from analysis of a typical display user's task. The model was used to interpret data from an experimental study, also based on this task.
    An Electronic Book: APTBook BIBA 513-519
      Mitsumasa Miyazawa; Kaoru Kinoshita; Minoru Kobayashi; Teruo Yokoyama; Yutaka Matsushita
    This paper describes an electronic book named APTBook in which we can leaf through the pages of it, furthermore we can either paste it with a memopad, or make a dog-ear, underlines. Since this APTBook is realized by the hierarchical data structure in which the more upper layer has the more rough information, the system can show many pages in a short time by using the animation of leafing through the pages. Furthermore, we can access a page in which a photograph is located in the upper-right corner by an access method similar to the memory structure of human being based on spatial location, nevertheless existing electronic media cannot manage it.
       Thus by, using APTBook, we can access the data without using database access based on keywords.
    "Good" Graphic Interfaces for "Good" Idea Organizers BIBA 521-526
      Kozo Sugiyama; Kazuo Misue
    "Good" graphic interfaces are indispensable for "good" idea organizers to arrange and organize lots of segments of information. We analyse the KJ Method, an idea organizing method famous in Japan, and consider graphic interface aspects of human-computer interactions. Some novel techniques such as automatic graph drawing, incremental editing, fisheye, diagram-document conversion etc. are presented. These techniques are useful also for browsers of hypertexts, collaborative work and so on.

    Detailed Design: User Support

    Help Systems: An Information-Sharing Approach BIBA 529-534
      Masayuki Kurisaki
    Although the knowledge-based approach has allowed the development of extensive help systems, it seems impossible to build a complete knowledge base for a large complex system such as an operating system, in advance. This paper describes the need for and the basic design of a help system which asks other users for help when it encounters an unknown situation. Only frequently asked questions and their solutions are stored in the initial state. However, a user can ask any question and is almost guaranteed to receive an answer even if it comes later by electronic mail. The collected solutions are kept in the knowledge base for future use. The system is intended to help non-expert programmers find a UNIX command or get advice from other users who belong to the same working group. The knowledge base of the system is regarded as an extension of the human memory shared by a group of users.
    End-User Dialogue Context Management of Office Automation Systems BIBAK 535-541
      Gang Lu; Claude Vanneste; Martin Ader
    Office work is characterized by concurrency and exception handling. While switching frequency back and forth between many concurrent activities, the office worker finds it difficult to manage dialogue context, (i.e., to restore quickly dialogue context regarding his current activity, manipulate numerous objects belonging to different activities on the limited size screen, coordinate the execution of all concurrent activities, and so on). Based on field observation we analyze the multiple activity characteristics of office work concerning dialogue context management. We postulate the requirements for an office automation system to assist end users in overcoming the difficulty of dialogue context management. Finally, we describe how a prototype office automation system IWS can partially meet these requirements, and propose future research directions.
    Keywords: User interface, Dialogue management, Office automation, End-user assistance, Multiple activity environment, Object-oriented systems
    Current Approaches & New Guidelines for Undo Support Design BIBA 543-548
      Yiya Yang
    Task-oriented commands cause essential steps in performing tasks within a system's scope whereas support-oriented commands inform a user about appropriate task-oriented commands, facilitate user-computer interactions or assure the integrity of a user's work. An undo capability is a support-oriented command facility which allows a user to reverse the effects of commands. It supports the fallible and fickle nature of users. In this paper major current undo support facilities are reviewed and critically compared. Design guidelines for undo support derived from a research project which prototyped sophisticated undo support in a widely used editing environment are then formulated.
    Interface Usability Engineering Under Practical Constraints: A Case Study in the Design of Undo Support BIBA 549-554
      Yiya Yang
    Employing usability engineering methods during interface design contributes to making the final product more usable. However, although classical evaluation methods are scientifically sound and can be effectively used for usability engineering, they are not practical, because system developers see them as too time consuming, expertise intensive and expensive to apply. Finding usability engineering methods, which can be applied under the practical constraints of time, cost and available expertise that normally shape interface development, is an important challenge for HCI. This paper discusses a project which shows how inexpensive and practical usability engineering methods can be employed during interface development to enhance its usability using the design of undo support as a case study.
    State versus History in User Interfaces BIBA 555-560
      W. B. Cowan; M. Wein
    Recent growth in window-oriented user interfaces with the implied evolution of interaction styles has led to two distinct approaches to user interfaces. The older style is terminal oriented and can be characterized as history-based: the current state of the system is inferred by the user from his or her knowledge of the sequence of previous commands. The newer window-oriented style is state-based in that the current state of the system is displayed in the window or the dialog box. Many interfaces attempt to combine elements from the two styles. This paper discusses the distinctive characteristics of each interface style, considering particularly user requirements that are inconsistent with the conceptual organization of the interface. We conclude that the two styles provide an almost non-overlapping set of capabilities, and that the choice of interface should be determined by the user's task requirements. The inevitable desire to encompass all capabilities in one interface demands significant generalization of the interface metaphor.
    The Personal Touch: A Study of User's Customization Practice BIBA 561-565
      Anker Helms Jorgensen; Allan Sauer
    Modern computer systems provide a rich variety of customization features that allow individual users to tailor the systems to their own needs. Although these features have existed for many years, very little is known about their actual use. This paper reports three studies of users' customization practice. The users ranged from fairly dp-naive professionals using pc's to experienced systems programmers using mainframe systems. About half of the experienced users applied the customization features and the other half had tried them, whereas the novice users had not used them at all. The changes made in the systems were mainly related to appearance, such as screen colour changed due to personal preferences or visual handicaps, and to task tailoring, such as redefinition of function keys and start-up functions. The satisficing principle explains the cases where the features are not used. The results indicate that user experience play an important role in customization practice.
    Inferring Task Structures from Interaction Protocols BIBA 567-572
      Franz Schiele; H. Ulrich Hoppe
    Powerful multi-purpose software may be used for an unforeseeable variety of tasks. To redesign, or to dynamically adapt such systems to the users' specific task requirements, new task structures must be inferred from the users' actual use of the system. An inductive method for inferring task schemata from protocols of user input has been explored, using different heuristics to filter out implausible task structures. An empirical study in the domain of file-handling tasks provides evidence for its ability to acquire meaningful task schemata.
    An Adaptive System Developer's Tool-Kit BIBA 573-577
      David Benyon; Dianne Murray; Frances Jennings
    Adaptive systems share some characteristics with other knowledge-based systems, but differ in other important respects. In particular, adaptive systems require a comprehensive model of the system users and have to make inferences not just about the domain, but also about the users' knowledge of the domain. This paper describes the design of a User Modelling Shell -- a system designed to meet the needs of adaptive system developers. The paper outlines the architecture of the system and the reasons for the chosen design and illustrates these principles with examples from an application of the system.

    Detailed Design: Hypermedia

    Roles for Tables of Contents as Hypertext Overviews BIBA 581-586
      T. T. Carey; W. T. Hunt; A. Lopez-Suarez
    Hypertext documents can provide increased access to information. However, users can experience disorientation as they jump between units in a document. This "lost in hypertext" phenomena is often thought of as a way-finding problem, to be solved by overviews of navigational structure. But we argue that it is often a problem of making sense of the document when the access sequence does not reflect its organizational structure, so that overviews must preserve and extend the user's organizational view.
       The paper describes a prototype overview aid, TableView, for users of hypertext systems. Extensions to traditional tables of contents allow TableView to integrate several roles for hypertext navigation aids. We discuss application of TableView for an online help system, and future extensions to incorporate additional overview roles.
    Navigation in Hypertext: A Critical Review of the Concept BIBA 587-592
      Andrew Dillon; Cliff McKnight; John Richardson
    With the advent of hypertext it has become widely accepted that the departure from the so-called "linear" structure of paper increases the likelihood of readers or users becoming lost. In this paper we will discuss this aspect of hypertext in terms of its validity, the lessons to be learned from the psychology of navigation and the applicability of the navigation metaphor to the hypertext domain.
    Combining Hypermedia Browsing with Formal Queries BIBA 593-598
      K.-H. Jerke; P. Szabo; A. Lesch; H. Rossler; T. Schwab; J. Herczeg
    This paper describes a system for retrieving and presenting multimedia objects (e.g. text, picture, graphic, video and audio) that are organized in an information network. The system combines the approaches of hypertext and formal query: at any time the user can express his intentions both by a formal query or navigate through the information space according to the hypertext paradigm.

    Detailed Design: Construction Tools

    An Experiment in Interactive Architectures BIBA 601-606
      Ernest Edmonds; Noriko Hagiwara
    The paper considers the interactive architecture known as the "Seeheim Model". A problem which emerged with that model as direct manipulation became more important is identified. A solution to the problem was proposed which involved the introduction of active objects in place of the application interface model. This proposal was evaluated by constructing a direct manipulation graphical interface using that architecture. Certain problems were encountered which were solved by modifying the architecture again. The paper describes the experiment and its results. A new interactive architecture is presented and its relationship to the source models demonstrated.
    SCENARIOO: A New Generation UIMS BIBA 607-612
      Brigitte Roudaud; Valerie Lavigne; Olivier Lagneau; Earl Minor
    UIMS technology and tools are still an evolving research area in the domain of human computer interaction. Although some techniques are today well defined, the user interface designer still has few tools to prototype, develop, debug, and assess a user interface. SCENARIOO [1] is a rich UIMS which provides many of these tools. SCENARIOO allows the user interface designer to prototype, develop, test and debug a user interface, either in a simulated or true application environment, with one or multiple running applications. It provides interactive graphical editors for development of each part of the user interface (presentation, dialogue control), and automatically generates the code of the final program. Thus, the interface designer can concentrate on good interface design, not programming details. Here we describe SCENARIOO.
    MUD: Multiple-View User Interface Design BIBAK 613-618
      David England
    The communication and visibility of information in a software engineering project is essential. In the user interface component of a project these needs are even more acute. This arises from the increasing underlying complexity of user interfaces and the multi-disciplinary nature of the teams required for their construction. This paper describes MUD, a Multiple-view User interface Design tool. This tool follows the hypertext paradigm by providing integrated, multiple instances of tools and browsers with which to view the attributes and relationships of a user interface. The base views include: end-user views, geometric views, structure views and object behaviour views.
    Keywords: User interface design, Rapid prototyping, Hypertext
    PENGUIN: A Language for Reactive Graphical User Interface Programming BIBA 619-624
      Sue-Ken Yap; Michael L. Scott
    PENGUIN is a grammar-based language for programming graphical user interfaces. Code for each thread of control in a multi-threaded application is confined to its own module, promoting modularity and reuse of code. Networks of PENGUIN components (each composed of an arbitrary number of modules) can be used to construct large reactive systems with parallel execution, internal protection boundaries, and plug-compatible communication interfaces. We argue that the PENGUIN building-block approach constitutes a more appropriate framework for user interface programming than the traditional Seeheim Model. We discuss the design of PENGUIN and relate our experiences with applications.
    Petri Net Objects for the Design, Validation and Prototyping of User-Driven Interfaces BIBAK 625-631
      Remi Bastide; Philippe Palanque
    Petri Net Objects (P.N.O.) are a high-level, object-structured dialect of Petri nets, primarily devised for the design of parallel systems. We show how this formalism can be used for the specification and design of event-driven interfaces, through the use of a real-life example. We then discuss the potential for dialogue validation, integration in UIMS and prototyping offered by this model.
    Keywords: User interface design, High level Petri nets, Prototyping, Validation, Object oriented software design
    An Object-Oriented UIMS for Rapid Prototyping BIBA 633-638
      Yen-Ping Shan
    User interface management systems (UIMSs) that support rapid prototyping often suffer from the limited range of interfaces that they can produce and the lack of support for the connection between the produced interface and its underlying application. This paper discusses a Mode Development Environment (MoDE) that addresses these problems.
    Do-It-Yourself Iconic Displays: Reconfigurable Iconic Representations of Application Objects BIBA 639-644
      Philip D. Gray; Kevin W. Waite; Stephen W. Draper
    It is often a major programming task to associate user interface objects with the application objects they are to represent. This discourages the exploration of alternative representations. In this paper we describe the architecture and interface of Iconographer, a system which enables iconic representations of (sets of) application objects to be specified in a highly interactive manner. At its heart is a direct manipulation "switchboard" by which attributes of application objects may be mapped on to icon attributes with a resultant change in the display.
    Localisation of Application Knowledge in Incremental Development of User Interfaces BIBA 645-650
      Philip D. Gray; Catherine A. Wood; Alistair C. Kilgour
    Localisation of application knowledge in user interface management systems may usefully be categorised with respect to two orthogonal dimensions: the horizontal dimension (the separation of application dependencies from dialogue control) and the vertical dimension (separation into task-related substructures).
       We discuss the relation of these issues to user interface development in the context of the Chimera User Interface Management System, which supports both dimensions of modularisation by a dynamically reconfigurable linkage component encapsulating local application knowledge.
    A UIMS for Knowledge Based Interface Template Generation and Interaction BIBA 651-657
      Christian Martin
    A knowledge based UIMS that exploits the semi-structured nature of office objects and tasks to generate flexible, user-adapted dialogue is presented. To produce interface templates for a broad range of interactive applications the conceptual structure of the application data and abstract dialogue objects serve as the elements of a frame-like representation of the user interface. This description is refined by a dialogue manager that evaluates rules for presentation and user preferences. A refined interface frame which is still independent of underlying I/O-tools, is mapped to the most suitable classes of the toolkit in the preferred environment (e.g. Andrew or OSF/Motif) by a generation component. The dialogue design process is illustrated for a multimedia editor application.
    Incorporating Metaphor in Automated Interface Design BIBA 659-664
      Brad Blumenthal
    Metaphoric interface design is a useful technique for making computer applications easier to learn and use. The MAID system uses a knowledge-based description of computer applications and real-world entities to automatically produce interface designs with metaphoric characteristics. MAID employs two strategies for producing metaphoric human interfaces: one imports characteristics such as appearance, relative size, etc. into the application, the other imports new objects suggested by the metaphor. MAID has been implemented and some results of its design runs are presented.

    Interactive Technologies and Techniques: Input

    The Role of Visual and Kinesthetic Feedback in the Prevention of Mode Errors BIBA 667-673
      Abigail J. Sellen; Gordon P. Kurtenbach; William A. S. Buxton
    The use of visual and kinesthetic feedback in preventing mode errors was investigated. Mode errors were defined in the context of text editing as attempting to issue navigation commands while in insert mode, or attempting to insert text while in command mode. Twelve novices and twelve expert users of the Unix-based text editor vi performed a simple text editing task in conjunction with a distractor task in four different conditions. These conditions consisted of comparing the use of keyboard versus foot pedal for changing mode, crossed with the presence or absence of visual feedback to indicate mode. Both visual and kinesthetic feedback were effective in reducing mode errors, although for experts visual feedback was redundant given that they were using a foot pedal. Other measures of system usability indicate the superiority of the use of a foot pedal over visual feedback in delivering system state information for this type of task.
    Windows on Tablets as a Means of Achieving Virtual Input Devices BIBA 675-681
      Ed Brown; William A. S. Buxton; Kevin Murtagh
    Users of computer systems are often constrained by the limited number of physical devices at their disposal. For displays, window systems have proven an effective way of addressing this problem. As commonly used, a window system partitions a single physical display into a number of different virtual displays. It is our objective to demonstrate that the model is also useful when applied to input.
       We show how the surface of a single input device, a tablet, can be partitioned into a number of virtual input devices. The demonstration makes a number of important points. First, it demonstrates that such usage can improve the power and flexibility of the user interfaces that we can implement with a given set of resources. Second, it demonstrates a property of tablets that distinguishes them from other input devices, such as mice. Third, it shows how the technique can be particularly effective when implemented using a touch sensitive tablet. And finally, it describes the implementation of a prototype an "input window manager" that greatly facilitates our ability to develop user interfaces using the technique.
       The research described has significant implications on direct manipulation interfaces, rapid prototyping, tailorability, and user interface management systems.
    Building Adaptive Interfaces with Neural Networks: The Glove-Talk Pilot Study BIBA 683-688
      S. Sidney Fels; Geoffrey E. Hinton
    A multilayer neural network can learn complicated mappings from inputs to outputs. After learning a mapping from a set of training examples, the network can generalize to new cases. Although the learning can be slow, the network runs extremely rapidly once it has learned so it can be used for real-time applications. To illustrate the potential of this technology for adaptive interfaces, we used a VPL DataGlove connected to a DECtalk speech synthesizer via five neural networks to implement a hand gesture to speech system. Using minor variations of the standard back-propagation learning procedure, the complex mapping of hand movements to speech is learned using data obtained from a single "speaker" in a simple training phase. With a 203 gesture-to-word vocabulary, the wrong word is produced less than 1% of the time, and no word is produced about 7% of the time. Adaptive control of the speaking rate and word stress is also available. The training times and final performance speed are improved by using small, separate networks for each naturally defined subtask. The system demonstrates that neural networks can be used to develop the complex mappings required in a high bandwidth interface that adapts to the individual user.
    FINGER -- A Language for Gesture Recognition BIBA 689-694
      Gerhard Weber
    We report on the development of a new kind of interaction for blind users. The interaction centers around our large, touch-sensitive tactile pin-matrix device. The fingers of the blind user are used for reading and simultaneously for specifying input in the form of gestures.
       We developed and implemented the language FINGER on a SUN workstation in C for describing gestures. Human-computer interaction based on finger-movements is thus formalized with FINGER. This formalization allows describing gestures in a textual notation, thereby providing a solid basis for their design. Furthermore, FINGER enables gestures to be adapted to the individual user by transformations of their textual representations.
    A Virtual Stereographic Pointer for a Real Three Dimensional Video World BIBA 695-700
      Paul Milgram; David Drascic; Julius Grodski
    A brief overview is given of a new display concept, involving superimposition of computer stereographic images onto a real world stereoscopic video display. The aim of the current system is to supply the user with a computer generated "virtual probe", for exploring, making measurements, and enhancing images within a real three dimensional video world. Development of the system is discussed, together with its capabilities and a number of practical considerations for its use. Although originally developed as an enhancement for telerobotic control only, use of the technology is predicted for a wide variety of novel multimedia applications.
    Force-to-Motion Functions for Pointing BIBA 701-706
      Joseph D. Rutledge; Ted Selker
    A pointing device which can be operated from typing position avoids time loss and distraction. We have built and investigated force-sensitive devices for this purpose. The critical link is the force-to-motion mapping. We have found principles which enable a force joystick to match the function and approach the performance of a mouse in pure pointing tasks, and to best it in mixed tasks, such as editing. Examples take into account task, user strategy and perceptual-motor limitations.
    Keyboard Layout for Occasional Users BIBA 707-712
      Nicolas Marmaras; Kostas Lyritzis
    The present study is a first attempt to solve the problem of keyboard layout for occasional users of computer-based systems. Sixty subjects, all customers of a department store, divided in four groups, have been asked to type a short phrase on specially tailored keyboards. The keyboard arrangements tested were the alphabetical-diagonal, the alphabetical-horizontal, the standard greek keyboard which is similar to the QWERTY layout, and a new alternative greek keyboard designed for typists. The obtained results showed that the alphabetical-diagonal arrangement achieved better typing speed rates, and that the differences between typing rates achieved with the other three layouts are practically insignificant.

    Interactive Technologies and Techniques: Output

    The Simulation of a Large Image Terminal Using Heath Robinson Techniques BIBA 715-720
      J. R. Harris; M. B. Harris; D. Th. Henskes
    A non-existent large image terminal with screen diagonal of one metre was simulated using various software and hardware techniques. Groups of non-computer scientists were used to evaluate its potential. This information is being used as vital input to the hardware design definition phase for a new generation of interactive terminals. The users were lecturers and students from a university languages department, and the information content was based on their normal classes. The software was developed on an Apple Macintosh IIX and the hardware was prepared from various items, including mirrors and string. We show that it is possible to have limited realistic interaction; to get a close proximity to the predicted visual effect; and to use real users in appropriate realistic scenarios.
    Evaluation of Flat Panel Display Properties on a High Fidelity Display Simulator BIBA 721-724
      Gerd Spenkelink; Henk Van Spijker; Ted White
    A dynamic high fidelity simulator of Flat Panel Displays (FPD's) was built in the Esprit project "Modelling and simulation of modern display technologies under office work conditions".
       Research methods using both experimental and subjective techniques have been developed and used for the purpose of simulator validation and the evaluation of FPD properties with respect to task performance and display acceptability for users.
       The research methods can provide a valuable contribution to a better understanding of display quality and the results may be used in the design of FPD's and display fonts as well as in the development of ergonomic standards and guidelines for these new display technologies. The methods are also suitable for research outside this specific application field.
    Colour Model Integration and Visualisation BIBA 725-728
      Peter A. Rhodes; M. Ronnier Luo; Stephen A. R. Scrivener
    Colour is an increasingly important element of human-computer interfaces and yet is difficult to control for a variety of reasons. In this paper, the most frequently used colour models are described and discussed. It is argued that no single model is going to be suitable for all applications. A system is proposed comprising a set of colour models that are integrated via mappings to and from the international standard colorimetric CIE XYZ System. The paper describes the models implemented to date in a system that allows interactive colour specification and communication.
    On the Visibility of Character Features on a VDU BIBA 729-734
      D. Bosman; T. N. White
    The legibility of characters depends on several factors, in the first place on the visibility of specific character features. Small character features may be difficult to perceive, even under supra-threshold conditions. Visibility of features is here taken as the probability that the corresponding amounts of light energy produce signals which can be interpreted by the brain. Therefore brightness, not luminance, determines perception. Early visual processing is modelled as a mapping of the character spatial luminance distribution into local brightness variations, measured in equal intervals. This is the basis for calculating visibility probability estimates of features of the presented characters and symbols. Influence on visibility is discussed of several common describing factors, such as average brightness, line separation, stroke width, active area and blur.
    Auditory Icons in Large-Scale Collaborative Environments BIBA 735-740
      William W. Gaver; Randall B. Smith
    We discuss the potential for auditory icons to address several common problems in large-scale, multiprocessing, and collaborative systems. These problems include those of confirming user-initiated actions, providing information about ongoing processes or system states, providing adequate navigational information, and signalling the existence and activity of other users who may be working in a part of the system that is not visible. We provide several examples of useful auditory icons drawn from a large, shared, multitasking environment called SharedARK, and discuss their implications for other systems.
    Interactive Scientific Visualization: An Assessment of a Virtual Reality System BIBA 741-745
      Philip J. Mercurio; Thomas D. Erickson
    A virtual reality system, consisting of a head-mounted stereoscopic display and a computer-interfaced glove, was assessed by examining interaction with a 3-D model of the human brain. Interactions were recorded on videotape. Non-trivial user interface issues were identified, ranging from constraints imposed by the nature of the wearable interface hardware, to the choice of gestures for controlling the interaction, to problems with a metaphor used in the interface. Some possible solutions are discussed. Sound solutions to these problems, coupled with increases in the computational power of the underlying hardware, are needed for virtual reality to realize its immense potential for scientific visualization.
    A Browser for Dynamic Multimedia Documents BIBA 747-751
      Suresh Anupindi
    A model for audio dominated multimedia documents is presented. Such documents are active and dynamic because of their audio content. However, revision of information is difficult because of the dynamic nature. We present static and dynamic scanning techniques which allow reviewing and describe a browser which employs these techniques. We also describe a method for customizing the presentation of dynamic documents using trails.

    Interactive Technologies and Techniques: Speech and Natural Language

    An Investigation into the Use of Error Recovery Dialogues in a User Interface Management System for Speech Recognition BIBA 755-760
      Mary Zajicek; Jill Hewitt
    Experiments were carried out to assess new users' attitudes to different versions of a speech input word processing system providing different error recovery strategies. While they preferred a simple error message to none at all, a more complex recovery dialogue lead to decreased satisfaction with the system. This paper describes the experiments carried out and explores possible reasons for the results.
    Feedback Requirements for Automatic Speech Recognition in Control Room Systems BIBA 761-766
      C. Baber; R. B. Stammers; R. G. Taylor
    Previous research into feedback requirements for users of ASR has tended to concentrate on verbal feedback: presented via text or via synthetic speech. In control room systems, auditory feedback is not viable for ASR, and textual feedback is potentially problematic. To counter these problems, feedback could be presented using symbols on existing displays. In this study, textual and symbolic feedback are compared for an error detection task, one of the prime uses of feedback for ASR. It was found that although performance using symbolic feedback did improve over time, it was significantly lower than that for textual feedback. Further, the type of symbol used also effected performance. Therefore for error detection at least, textual feedback is preferable to symbolic feedback. However, the role of symbolic feedback in more direct task control is yet to be explored.
    Spoken Language Interaction in a Spreadsheet Task BIBA 767-772
      Alexander I. Rudnicky; Michelle Sakamoto; Joseph H. Polifroni
    To study the spoken language interface in the context of a complex problem-solving task, we had a group of users perform a spreadsheet task, alternating voice and keyboard input. A total of 40 tasks were performed by each participant, the first thirty in a group (over several days), the remaining ones a month later. The voice spreadsheet program used in this study was extensively instrumented to provide detailed information about the components of the interaction. These data, as well as analysis of the participants' utterances and recognizer output, provide a fairly detailed picture of spoken language interaction.
    Case Study of Development of a User Interface for a Voice Activated Dialing Service BIBA 773-777
      Deborah Lawrence; Rory Stuart
    A user interface for a Voice Activated Dialing service was designed and evaluated. The service will use speaker dependent recognition technology based in the telephone network to allow users to place a call by saying a name into a regular touch tone or rotary telephone. The interface was designed iteratively in three design-evaluation cycles, each with 20 subjects selected to represent the population of adult telephone users. Evaluations examined user errors, task completion, names used, recognition performance, responses to Likert-like questionnaires rating the interface, and responses to open-ended questions. A comparison was made of design variations (alternative menus, access methods, and voice- vs. tone-prompts), and of usability of the interface with different types of telephones. Subjects had little difficulty understanding and following the prompted procedures for adding, erasing, reviewing and voice-dialing names. Few errors were common across subjects. However there were several types of user difficulty in interacting with the recognizer, and the human factors of human-recognizer interaction is discussed.
    A Voice Recognition Interface for a Telecommunications Basic Business Group Attendant Console BIBA 779-785
      Ismail Sola; Don Shepard
    The Human Machine Interface (HMI) Group at NEC America has designed a voice recognition interface for the Basic Business Group (BBG) Attendant Console. The personal computer (PC) based console provides call processing services to telephone operating company Business Group customers. Voice recognized commands have been implemented to facilitate more effective and efficient call processing. An experimental analysis is presented which shows the effectiveness of the voice interface in this product.
    Observations on Using Speech Input for Window Navigation BIBA 787-793
      Chris Schmandt; Debby Hindus; Mark S. Ackerman; Sanjay Manandhar
    We discuss the suitability of speech recognition for navigating within a window system and we describe Xspeak, an implementation of voice control for the X Window System. We made this interface available to a number of student programmers, and compared the use of speech and a pointer for window navigation through empirical and observational means. Our experience indicates that speech was attractive for some users, and we comment on their activities and recognition accuracy. These observations reveal pitfalls and advantages of using speech input in windows systems.
    The Design and Implementation of a Context Sensitive Natural Language Interface to Management Information BIBA 795-800
      Alan Burton; Anthony P. Steward
    Natural Language Interfaces (NLIs) make database (DB) query easier for infrequent computers. Interfaces to management information are particularly problematical because of the unpredictability of queries. This paper describes ATMI a context sensitive NLI to a working Oracle DB in a managerial environment. We discuss why context free grammars cannot provide an adequate model of the range of English which needs to be covered. A knowledge based approach which enables pronouns and other ambiguous terms to be resolved intelligently is described. Our current system demonstrates the feasibility of this approach.
    Recent Approaches to Natural Language Generation BIBA 801-805
      Lee Feder
    An HCI system can produce speech or text output if it is coupled with a language generation component. Such a module should be capable of producing reliable text that uses the resources of the language to produce the best description of the data in the current context. Past language generators have been criticised on several counts, and developments are underway to improve the situation. This paper examines the criticisms, and their implications for HCI systems. We look at the new approaches, and present our own solutions.

    Applications and Case Studies: Knowledge-Based Systems

    User Centered Explanations in Knowledge Based Systems BIBA 809-814
      Klemens Waldhor; Hans Anschutz
    One advantage of expert systems is the ability to explain its reasoning process. Many commercial systems support explaining only at rule trace level. Knowledge engineers understand such explanations; the everyday user who is not expert in that area is confused. Our approach allows the definition of user centered explanations and justifications by knowledge engineers. These explanations and justifications of conclusions are adapted to the way the user solves problems by cutting off irrelevant details and presenting only main steps in the reasoning process. A sophisticated user interface allows the user to ask questions about the system behaviour in various ways.
    Intelligent User Interface for a Conventional Program BIBA 815-820
      J. Junger; G. Bouma
    In this paper we describe an intelligent user interface to CHEMSIM, a mathematical simulation program. The interface is added to the already existing and completed program as a separate front-end. In this paper three aspects of the interface will be described: the task model represented as a hierarchical tree, the use of the tree for knowledge-representation and the graphical component. The choice of an object-oriented approach enables including in the class definitions both task aspects and domain knowledge aspects, and this way integrating the two. This approach simplifies the architecture as well. In addition, the same hierarchical tree representation is used also for the graphical component consisting of windows, menus and icons. The dynamic icons, governed by rules give the user direct feedback on his/her actions and help.
    Knowledge Acquisition and Hypertext in Manufacturing BIBA 821-826
      S. M. Hajsadr; A. P. Steward
    The particular nature of manufacturing knowledge is proposed and the demands that this nature places on a company are outlined. To respond to this situation many companies have instituted problem solving teams that gather, organise and consolidate worthwhile knowledge. To assist this process in a particular company we have designed two software tools; one is based on an expert system shell and the other on hypertext. These tools are available to machine operators in the company to browse or to contribute to the knowledge. The forms of these tools are described and the advantages of hypertext as being closer to production knowledge is discussed. We finally suggest that the benefits of the rule-based approach can be embedded in hypertext.
    Knowledge Based User Interfaces for Scientific Programs BIBA 827-832
      Henk J. Van Zuylen; Herman Gerritsen
    In a pilot study, a knowledge based support system has been developed for a scientific computer program. Some interesting phenomena have been observed during this process. The need for a non-linear development process could be shown. The development of a knowledge based support system for this complicated application program resulted in a specification of a system that could be considered as a user interface. The approach to use knowledge elicitation to get a user interface design showed to have some useful possibilities.

    Applications and Case Studies: Computer Supported Co-operative Work

    Concurrent Editing: The Groups Interface BIBA 835-840
      Judith S. Olson; Gary M. Olson; Lisbeth A. Mack; Pierre Wellner
    We review aspects of systems built for group work that allow real-time, concurrent editing of a single work object. Existing systems vary in both what group functions they offer users (e.g., whether simultaneous editing is possible or it must proceed one by one) and how these functions appear in the user interface (e.g. what signals are given to the user that the window is public or private). Design alternatives suggested by existing systems are analyzed in terms of their value for various phases of group work and their support for individuals' needs in coordinating their work.
    Characteristics of Well-Designed Electronic Communications Systems BIBA 841-847
      Patrick A. Holleran; Richard W. Haller
    Software designed to facilitate communication among people has tremendous potential to improve the quality and consequences of human cooperative effort. The success of integrated electronic communications systems depends on both the usability and functionality of the software. The present paper briefly discusses both general principles of software design and those specifically relevant to communications systems. It offers recommendations for the design of communications systems which will optimize the accessibility, usability and utility of this software.
    Process Modelling and CSCW: An Application of IPSE Technology to Medical Office Work BIBA 849-852
      Janet Maresh; David Wastell
    An Integrated Process Support Environment (IPSE) refers to a work environment (typically software development) in which the cooperative activities of men and machines are coordinated by a computer system according to a formally defined schema. In this paper we demonstrate the use of the process modelling language PML (developed in the Alvey-supported IPSE 2.5 project) to represent the cooperative structure of work underlying the administration of hospital outpatients activity. The semantics of PML and the methodology of process modelling are discussed and consideration is given to the implications of IPSE technology for the design of office systems.
    Tools that Support Human-Human Communication in the Automated Office BIBA 853-859
      I. D. Benest; D. Dukic
    This paper reports on the progress made so far on the development of a new office automation environment. Such an environment must support the technical access to the very large quantities of information available in an office. It must also provide instinctive mechanisms that offer both casual and formal access to that information. It attempts to widen the keyhole effect that is present in other office information systems, and also imposes a surreptitiously managed work environment. This contribution describes a set of tools (for example electronic mail and computer conferencing) developed to support human-human cooperative work in this automated office.

    Applications and Case Studies: Applications

    Smartwriter: A Tool-Based Wordprocessor for Adult Literacy Students BIBA 863-868
      David Ellis; Jane Horton; Philip Black
    This paper describes the design and development of a computer-based Literacy Tutor centred on a mouse-driven wordprocessor. The appropriateness of the design for literacy students is emphasised; the design objectives being simplicity and consistency rather than the implementation of a large collection of 'interesting' features. A novel tool-based approach has been adopted for the wordprocessor and associated learning modules. The suitability of using an object-oriented design methodology for this research project is clearly demonstrated. The subsequent development of the system for profoundly deaf students further exposes the advantages of this approach. Both systems have been evaluated in the field.
    The Interface to a Hypertext Journal BIBA 869-874
      Annette Simpson
    This paper reports some of the findings of a study comparing two interfaces to a hypertext database of academic journal articles. The design of the interface to the individual articles was derived from the results of previous studies by the author. The aim of the present investigation was to determine whether those features found to assist readers in navigating through, and extracting information from, individual journal articles were of benefit when applied to the interface to a database of such articles. The influence of interface type on readers' ability to locate the information required to answer two essay-type questions was examined. Performance, both in terms of the amount of information located and the efficiency with which it was located, was significantly better when using an interface containing these features as compared to an interface whose features were derived from the paper medium.
    A Fisheye Presentation Strategy: Aircraft Maintenance Data BIBA 875-880
      Deborah A. Mitta
    A computer interface that presents information in a usable format will typically improve the quality of human-computer interaction (HCI). One presentation technique recently developed as a means of enhancing the quality of HCI is known as the fisheye lens viewing strategy. This paper will discuss an application of the fisheye lens viewing strategy to the presentation of aircraft maintenance data. The research results presented in this paper will demonstrate that the fisheye strategy, through its ability to prioritize interface information, can be used as a mechanism for filtering details of maintenance data.
    Supporting Exploratory Learning BIBA 881-885
      Andrew Howes; Stephen J. Payne
    This paper introduces the notion of a Learning Support Environment (LSE), which is a collection of tools designed to support exploratory learning of computer applications. An implementation of an LSE for a particular interactive device is motivated in terms of an analysis of the cognitive problems faced by the exploratory learner. The implementation includes four support tools; a Task-action Trace, a Metaphor Micro-world, an Animation Machine and a Buddy Learner.
    Application of Cognitive Modeling and Knowledge Measurement in Diagnosis and Training of Complex Skills BIBA 887-892
      Yan M. Yufik
    We discuss an innovative approach to the design of intelligent training systems (ITS), integrating methods of Artificial Intelligence, Cognitive Modeling and Hypermedia New ITS will facilitate personalized training in complex technical domains, and will support a. acquisition of expert domain models, b. simulation of expert strategies of models manipulation, c. analysis and visualization of models structure, and d. quantitative comparison of students and expert models. The advantage of the proposed approach is the underlying semiformal format of domain knowledge representation which is both cognitively compatible and computer executable. The paper starts by defining the problem of knowledge measurement and transfer in complex interactive tasks, and then discusses ITS architecture and training methods.

    Applications and Case Studies: Software Development

    Software Reusability: Delivering Productivity Gains or Short Cuts BIBA 895-901
      Alistair Sutcliffe; Neil Maiden
    It has been claimed that software reuse can produce considerable productivity gains in system development. Although much software engineering research has been undertaken to deliver reusability in CASE tools there is little knowledge about how system developers actual reuse specifications. This paper reports practical research into reuse scenarios based on providing analogous specifications for systems analysts. The results are encouraging for reuse but caution that reuse may also create mistakes in specifications by erroneous transfer of knowledge.
    A Project-Orientated View of CSCW BIBA 903-908
      N. R. Seel; G. N. Gilbert; M. E. Morris
    Project Support Environments (PSEs), a type of computer supported cooperative work (CSCW) system, are examined in terms of the management forms which their designs appear to assume. Data-oriented PSEs can be seen as aiming to support management by direct authority. Process-oriented PSEs are aligned to the exercise of technical authority. However, much of the work of the 'professional communities' found in organisations is conducted, not under either of these two more traditional forms of management, but under a pluralist form, in which professionals are given a degree of responsible autonomy. Neither data- nor process-oriented PSEs are well suited to this management form. The characteristics of a PSE designed to support professional communities are outlined. Such a PSE, a 'Professional Community Support' system, needs to provide resources for negotiated cooperation in order to support what is argued to be the key activity of professionals -- issue handling. These resources must be designed to support explicitly both globally and locally managed interactions, with it being possible to embed either within the other.
    Satisfying the Need to Know: Interpersonal Information Access BIBA 909-915
      Robert E. Kraut; Lynn A. Streeter
    We examine the ability of traditional and computer-based communication technologies to spread organizational and task knowledge in large scale software development environments. It is our contention that the principal problems in software development are social and organizational, rather than cognitive. We review: (1) factors that make improving software development formidable, and (2) technological aids and project management methods that have been tried as possible "solutions," (3) a survey we are conducting on coordination techniques in large projects and conclude by (4) discussing candidate information/communication technologies to support coordination.
    ConversationBuilder: An Open Architecture for Collaborative Work BIBA 917-922
      Simon M. Kaplan
    Software process support tools of necessity be highly tailorable to mesh with the culture of, and tools used by, groups of programmers. They must also support the activities of groups in a natural and integrated manner. The ConversationBuilder is an 'open' tool in which provides support for cooperative, goal-directed group activities such as the software process.

    Applications and Case Studies: Programming

    Learning to Program in Another Language BIBA 925-930
      Jean Scholtz; Susan Wiedenbeck
    Our objective in this study was to examine how programmers go about learning new programming languages and to identify which areas of program development presented difficulties. Verbalizations from think aloud protocols were classified as one of five kinds of knowledge: syntax, semantics, strategic planning, tactical planning, and implementation planning. Implementation planning occupied over 50 percent of subjects' efforts, irrespective of language and level of expertise. Difficulties in a dissimilar language were due to subjects' failure to construct appropriate tactical plans. In a similar language difficulties centered on subjects' inability to locate appropriate constructs.
    {Upsi}πADAPTερ -- Individualizing Hypertext BIBA 931-936
      Heinz-Dieter Bocker; Hubertus Hohl; Thomas Schwab
    {Upsi}πADAPTερ is an adaptive Hypertext system that offers individualized access to and presentations of tutorial information. Depending on the user's current knowledge state which is represented in a dynamic user model, it identifies and suggests nodes of the network of tutorial topics which may serve as starting points for a succeeding browsing session. Within this browser, the presentation of tutorial information is determined by the data contained in the user model. During the browsing session, the sequence of knowledge units explored by the user is analyzed to infer the user's current knowledge state and maintain the user model.
    Minimalist Planning Tools in an Instructional System for Smalltalk Programming BIBA 937-944
      Mark K. Singley; John M. Carroll
    We describe the design of an instructional system for Smalltalk that attempts to reify students' goals and plans through the application of minimalist planning dialogues. Minimalist planning dialogues are sparse in that the planning knowledge is not stated explicitly but rather is imbedded into the structure of a planning "tool." They are situated in that the planning tools are fully integrated into the physical problem space and in essence provide an elaborated view of that space. We contrast our system with other current attempts at reifying students' goals and plans and sketch out problems and prospects for future work.
    Why Program Comprehension is (or is Not) Affected by Surface Features BIBA 945-950
      Barbee T. Mynatt
    The literature contains a variety of conflicting reports on the effect of various surface features such as variable naming style, indenting and commenting on program comprehension. In some cases these features appear to aid comprehension as intended, and in other cases they do not. Studies done at Bowling Green State University, reported here, have likewise shown conflicting outcomes. Pennington's (1987) model of programmer comprehension based on text structure knowledge is used to reconcile many of the results. According to her model different levels of knowledge, ranging from operational to functional, are extracted during comprehension. It was hypothesized that poor variable names would affect a programmer's comprehension of function, but would not affect the other sorts of knowledge. An experiment comparing comprehension of programs using either meaningful or nonsense variable names found the predicted result. The implications of the results vis-a-vis the true effects of surface features are discussed.
    The Generalized Unification Parser: Modelling the Parsing of Notations BIBA 951-957
      T. R. G. Green; Andrea Borning
    Experienced readers of notations pick out structures such as 'plans' or 'cliches'. It has been claimed that these are easier to perceive in some notations than others, making some notations more 'role-expressive'. We present a computational model of parsing, derived from a natural language parsing model, which has been applied to programming cliches and is capable of using typographical features as parsing aids. Preliminary results suggest that languages where cliche-structures are marked by lexical key-words (e.g. the Pascal family) are easier to parse than languages where cliche-structures are not marked by surface cues (e.g. Prolog), and more significantly, that the difference will increase rapidly with the complexity of the program. Experiments are in progress to test this prediction, which implies that support tools for illuminating program structure may be needed.
    Program Comprehension Beyond the Line BIBA 959-963
      Scott P. Robertson; Erle F. Davis
    Comprehension of computer program code has often been compared with text comprehension. We argue, though, that the requirements of code comprehension make it more of a problem-solving task that happens to use text-like material. We present data on search patterns and reading times in code comprehension that support this view. Specifically, we found that programmers examine code in repeated cycles that cover functionally relevant units. We suggest some problem-solving goals that guide search through code and show that line scanning times vary with hypothesized problem-solving activities. In a direct comparison of programmers reading isolated lines versus lines in the context of program comprehension we show that a simple model of microstructure parsing predicts reading times better for isolated lines than for lines in the context of a program.
    Expert Programmers Re-Establish Intentions when Debugging Another Programmer's Program BIBA 965-970
      Ray Waddington; Roger Henry
    When discussing software debugging, some authors have discussed the experience of debugging another programmer's program as being somehow different from that of debugging one's own program. Software psychologists attempt to understand the nature of debugging expertise, but have ignored the potentially fruitful method of looking empirically at the differences between debugging in the two situations. We discuss a model of debugging expertise which addresses the relationship of program authorship to debugging strategy. This model predicts that when debugging another programmer's program, experts will use a strategy of re-establishing the original author's intentions in order to debug it. We report an experiment, conducted with expert programmers, which supports this prediction. We also discuss the implications of this result for the design of debugging aids to support expert programmers when they are debugging another programmer's program.
    Difficulties in Designing with an Object-Oriented Language: An Empirical Study BIBA 971-976
      Francoise Detienne
    An experiment has been conducted to study the activity of program design developed by programmers experienced in classical procedural languages as they use an object-oriented programming (OOP) language. This paper focuses on the analysis of the difficulties programmers experienced in designing with OOP language. An important difficulty is to articulate the declarative and the procedural characteristics of the solution. This study highlights the importance of a representation of the procedure so as to construct the static relations between objects. This result does not support the hypothesis on naturalness of design with an OOP language made by advocates of OOP. This experiment also show that previous knowledge of programming languages may produce negative effects in the acquisition of a new language.
    The Spreadsheet Interface: A Basis for End User Programming BIBA 977-983
      Bonnie A. Nardi; James R. Miller
    This paper describes the properties of the spreadsheet interface and the ways in which spread-sheets support users with little or no formal training in programming. We analyze the spreadsheet formula language through which users express mathematical relations and the tabular grid which permits users to view, structure and display data. Based on our analysis of the formula language and the tabular grid, we argue that user programming environments should be characterized by (1) a limited set of carefully chosen, high-level, task-specific operations that are sufficient for building applications within a restricted domain, and (2) a strong visual format for structuring and presenting data.
    Action Representation for Home Automation BIBA 985-990
      Suzanne Sebillotte
    A study on the representation of the actions in the context of programming various home devices (oven, TV, heating etc.) is reported. Previous studies have shown that: 1) in a work situation, subjects' representations of their task are based on hierarchical levels of abstraction, 2) in the context of programming home devices subjects referred to general concepts of activities (e.g. "tuning"). Two experiments, inspired by those of Galambos (1986) were conducted in order to specify more fully these general concepts. Result showed that these concepts were deeply rooted in subjects' representations, independently of the devices. Implications of these results for home interface design are described, concerning especially command naming.
    Browsing Through Program Execution BIBA 991-996
      Heinz-Dieter Bocker; Jurgen Herczeg
    The system TRACK is a trace component for animation and debugging of SMALLTALK-80 programs Unlike traditional tracers it is designed as a construction kit utilizing interaction techniques based on graphical visualization and direct manipulation. A trace is specified by manipulating graphical objects in a trace window. Different trace windows may provide insight into different parts of a program. This paper describes how TRACK interacts with the standard programming tools of SMALLTALK-80 (class browsers, inspectors, debuggers, etc.) and how tracing and browsing techniques are combined to visualize the execution of a program.
    Compressing and Comparing Metric Execution Spaces BIBA 997-1002
      John Domingue
    To help programmers evaluate the efficiency of their code during debugging, we are developing the Transparent Rule Interpreter Monitoring System (TRIMS). This provides the user with a visual representation of both behavioural and performance aspects of rule-based programs. Up to now, visualization within TRIMS has been applied only to the qualitative behaviour of a program. This paper describes the recent incorporation of metric information into the system, enabling the programmer to visualize the time various parts of the program take to run. The use of metric information is described in relation to two facilities currently available in TRIMS: compression and comparison. Compression enables the programmer to hide away parts of the execution, permitting the visualization of arbitrarily large execution spaces. The comparison of similar execution spaces can answer such questions as, 'Why does my program run so slowly with this particular working memory set?'

    Doctoral Programme

    A Psychology of Programming for Design BIB 1005-1006
      Rachel K. E. Bellamy
    Cognitive Style and Intelligent Help BIB 1007-1008
      Lynne Coventry
    Support for Understanding and Participation in a Distributed Problem Solving System BIB 1009-1010
      C. M. Duursma
    The Role of Analogy in Training Computer Users BIBA 1011-1012
      Jonathan Elcock
    This paper summarises work on training by analogies. Four studies are discussed. The conclusions drawn from these studies is that while analogies have an effect in training that effect is not simply beneficial. Theoretical issues arising from these studies are then discussed.
    Linguistic Models in the Design of Cooperative Help Systems BIB 1013-1014
      Charles Elliot
    An Environment to Support the Use of Program Examples while Learning to Program in LISP BIB 1015-1016
      Katherine Wanjiru Getao
    Modelling Cognitive Aspects of Complex Control Tasks BIB 1017-1018
      Simon Grant
    Using Temporal Logic to Prototype Interactive Systems BIBK 1019-1020
      C. W. Johnson
    Keywords: Formal methods, Prototyping, Rendering, Temporal logic
    A Development Environment for the Design of Multimodal Colourgraphic Human-Computer Interfaces BIBA 1021-1024
      M. Langen; G. Rau
    The following object-oriented concept improves the design procedure of complex human-computer interfaces by a suitable development environment. This environment supports the process of evolutionary prototyping, i.e. prototypes are iteratively modified until a satisfying version is achieved. In addition to conventional approaches, this development environment integrates software tools for multimodal interaction and interactive colour manipulation. The application of this environment is demonstrated in the design of the human-computer interface of an anesthesia expert assist system.
    Advanced User Interfaces for Distributed Group Communication BIBA 1025-1027
      Leandro Navarro
    The aim of this work is to define an architecture framework of User Interfaces (UI) for Group Communication (GC) activities. After a brief presentation of the background and objectives of this work we will discuss several topics in the scenario of human to group interaction processes. The resulting UI model is based on the integration of several technologies and components: object orientation, multimedia, intelligent agents, UI languages.
    Logic Descriptions in Rapid Prototyping of Applications BIBA 1029-1030
      Lars Oestreicher
    This paper presents a description method for design of new systems, using formal descriptions combined with rapid prototyping. The basic formalism is First Order Predicate Logic in the shape of Horn Clauses. The descriptions are used as conceptual models of the functionality, formed with main input from task analysis of the user's expected work tasks with the system. The possibility to immediately execute the description as an executable specification of the new system also brings this research result into the area of rapid prototyping of new designs.
    Graphical Treatment of Natural Language in HCI BIBA 1031-1032
      Ronald A. Singer
    This paper discusses on-going research which indicates that graphical interfaces can offer users a more effective means of communicating their intentions to the system than is possible with NL. The relationships between user thoughts and graphical objects, must if they are to be natural and effective, reflect the structure of human discourse. This requires an interface which can understand the relation of subsequent thoughts to preceding ones. Circuit I (Singer, 1989), an object-oriented prototype has clearly demonstrated that the notion of embedding discourse phenomena (anaphora and ellipsis) as handled by SOPHIE (Brown, Burton, et al, 1982) within a graphics environment is a viable alternative to that of NL given the current unresolved problems. A small scale evaluation of the prototype has been carried out, and preliminary findings have been very encouraging.
    Run Time Interface Specification, Using Direct Manipulation BIBA 1033-1034
      Robert Tibbitt-Eggleton
    A brief overview is given of a prototype User Interface Management System (UIMS) that is being produced as part of a research degree. One of the main features of this UIMS is the ability to create/alter the interface to an application as the application runs.
    Learning a Word Processing Task: About Documentation, Help and Task Complexity BIBA 1035-1036
      Anne Van Laethem
    This work is about learning to use a personal computer for a task of medium complexity.

    Panel Sessions

    HCI Seen from the Perspective of Software Developers BIB 1039-1042
      John L. Bennett; Peter Conklin; Karmen Guevara; Wendy Mackay; Tom Sancha
    User Participation in HCI Research -- Effects on Processes and Results BIB 1043-1046
      Yvonne Wærn; Liam Bannon; Toomas Timpka; Werner Schneider
    Interactively Supporting the Software Process BIB 1047-1048
      Simon M. Kaplan; Anthony Finkelstein; Gail Kaiser; Kevin Ryan; Wilhelm Schafer
    Task Analysis: The Oft Missing Step in the Development of Computer-Human Interfaces; Its Desirable Nature, Value, and Role BIB 1051-1054
      Richard I. Anderson; John M. Carroll; John F. McGrew; Jonathan Grudin; Dominique L. Scapin
    New Approaches to Theory in HCI: How Should We Judge Their Acceptability? BIB 1055-1058
      Andrew Monk; John Carroll; Michael Harrison; John Long; Richard Young
    Multi-Agent Interaction BIB 1059-1061
      Nigel Seel; Julia Galliers; George Kiss; Stephen Scrivener
    Multi-Dimensional Interfaces for Software Design BIB 1063-1066
      Tim Dudley; Ronald Baecker; Marc Eisenstadt; Ephraim Glinert; Mary Beth Rosson
    Usability Engineering on a Budget BIBA 1067-1070
      Jakob Nielsen; Susan M. Dray; James D. Foley; Paul Walsh; Peter Wright
    This panel will discuss how to get the "most bang for the buck" in usability engineering. What should one do when the budget is restricted and it is impossible to do everything by the book? How can one introduce usability methods in companies that currently have no systematic usability efforts?