HCI Bibliography Home | HCI Conferences | BCSHCI Archive | Detailed Records | RefWorks | EndNote | Hide Abstracts
BCSHCI Tables of Contents: 8586878889919293949596979800

Proceedings of the HCI'88 Conference on People and Computers IV

Fullname:Proceedings of the Fourth Conference of the British Computer Society Human Computer Interaction Specialist Group
Note:People and Computers IV
Editors:Dylan M. Jones; Russel Winder
Location:University of Manchester
Dates:1988-Sep-05 to 1988-May-09
Publisher:Cambridge University Press
Standard No:ISBN 0-521-36553-8; hcibib: BCSHCI88
Papers:36
Pages:594
  1. Analysis for Early Design
  2. Systems Design: Methods, Tools and Processes
  3. Implementation Tools
  4. Systems Support
  5. Input/Output
  6. Dialogue Design
  7. Artificial Intelligence Issues
Preface BIBA 3-5
  Martin Thomas
In the UK there is one annual national conference on HCI, organised by the British Computer Society's HCI Specialist Group. This book contains the refereed papers presented at the HCI'88 conference, held in September 1988 at UMIST, Manchester, England. This preface sets out to put the contents into context, both for people reading this book as it stands, and those attending the conference.
Computers for the People: HCI in Prospect. An Introduction to the HCI'88 Conference Proceedings BIB 7-10
  Dylan Jones

Analysis for Early Design

Implications of Current Design Practice for the Use of HCI Techniques BIBAK 13-34
  Victoria Bellotti
A study of commercial system-interface design projects was carried out in order to determine the nature of real world design practice. Of particular interest were two questions; the first being whether commercial design makes use of HCI design and evaluative techniques, and the second being whether commercial design satisfies the requirements for successful application of these design aids. The findings suggested that commercial design practice varies both in the constraints under which it operates, and in the approaches adopted. Although many problems relating to interface design appear to be tractable to HCI techniques, these techniques are rarely used. Conditions in commercial design practice sometimes act as unavoidable constraints on what designers can do. These constraints have important implications for the applicability, or inapplicability, of HCI design and evaluative techniques.
Keywords: Design practice, HCI task analysis techniques
Task-Related Knowledge Structures: Analysis, Modelling and Application BIBAK 35-62
  Peter Johnson; Hilary Johnson; Ray Waddington; Alan Shouls
A theoretical and methodological approach to task modelling is described, with a worked example of the resultant model. The theory holds that task knowledge is represented in a person's memory and that this knowledge can be described by a Task Knowledge Structure (TKS). The method of analysis has been developed for carrying out analyses of real world tasks. The method uses a variety of techniques for collecting information about task knowledge. A second perspective of the paper shows how a developed TKS model can be decomposed into a design for a software system to support the identified tasks within the domain of the analysis. This decompositional method uses the structure of frames to provide consistency between different levels of design decomposition.
Keywords: Task modelling, Knowledge representation, Frames, Design decomposition
Abstract, Generic Models of Interactive Systems BIBAK 63-77
  Alan Dix
For several years at York, we have been investigating the use of abstract models in the design of interactive systems. I will describe why we originally pursued this line and the benefits that have ensued. I will only briefly describe specific models as examples where appropriate. There is an underlying assumption that formal methods are being used during the software design process, but the analysis proves useful even when this is not the case.
Keywords: Formal methods, Design principles, Requirements capture
Analysing the Scope of Cognitive Models in Human-Computer Interaction: A Trade-Off Approach BIBAK 79-93
  Tony Simon
One of the main contributions of Cognitive Science to HCI has been the development of predictive models of user behaviour. However, such models are necessarily limited in the scope of predictions they can make; their strengths usually being determined on the basis of pragmatic trade-offs. At present, no rational taxonomy of the different types of model exists. Thus, would be user-modellers find little guidance about which model is most likely to deliver the kind of predictions in which they are interested. Even less available is information about what will not be delivered when employing any given model. This paper presents a representation of the space of some user-models in HCI which reveals their scope by making explicit such trade-offs.
Keywords: User-modelling, Cognitive science

Systems Design: Methods, Tools and Processes

The Design and Evaluation of an Animated Programming Environment BIBAK 97-109
  Kaizad B. Heerjee; Michael T. Swanston; Colin J. Miller; William B. Samson
APE, an Animated Programming Environment, is an interactive, graphical, program design and development system, that embodies structured programming and top-down design. The system supports the development of programs for a variety of block structured languages whilst working conceptually at the level of Jackson diagrams. Evaluation of APE has been carried out during the design and implementation stages of the development life-cycle. The evaluation was based on responses to a questionnaire and a comparison with conventional means of generating code. The questionnaire evaluation elicited users' general impressions about the system and its interface, and their detailed views on more specific aspects of the system. The comparative evaluation showed no difference in the mean quality of the solution to a programming problem, but a significantly reduced variance in quality compared to conventional methods.
Keywords: Human-computer interface, Software evaluation, Programming environments
Overcoming Obstacles to the Validation of User Requirements Specifications BIBAK 111-122
  M. A. R. Kirby; C. J. H. Fowler; L. A. Macaulay
Poor specification of user requirements is a major reason why computer systems fail or dysfunction. One way of addressing this problem is to validate User Requirements Specifications before proceeding with system development. To date, it has only been possible to validate specifications against checklists of what they should contain. This type of validation indicates gaps but does not check the reliability of a specification; nor does it explain the implications of specification deficiencies for performance of the finished product.
   This paper identifies obstacles to the development of validation techniques that do check reliability and do explain the implications of specification deficiencies. An approach to overcoming these obstacles is discussed, particularly a method for ensuring that a specification is verified with the right set of users, and a method of manipulating and analysing the information in a specification to predict dysfunction. This approach has been used to develop the Specified User Requirements Validation and Explication (SURVE) technique.
Keywords: Requirements specification, User requirements, System validation
The Representation of User Interface Style BIBAK 123-143
  William M. Newman
This paper identifies the need for representations of styles of user interface, particularly as a basis for choosing an application style or porting an application to a new environment. It identifies the requirements that a style representation should meet, and then proceeds to develop a representation based on the use of points of style. It shows how this representation is capable of relating style to user requirements, how it helps construct the style's underlying argument, and how it exposes stylistic weaknesses. Several worked examples are included.
Keywords: Human computer interaction, User interface design, Design style, Design rationale, Requirements, Applications design, Software environments, User interface toolkits, Apple Macintosh
Some Experiences in Integrating Specification of Human Computer Interaction within a Structured System Development Method BIBAK 145-160
  Alistair Sutcliffe
Procedures for integrating task analysis and design of human computer interfaces into a structured system design method, Jackson system development (JSD) are described. JSD process structure diagrams are used to describe tasks which are then evaluated for cognitive complexity. Task allocation and complexity analysis produced specification of human tasks, highlighted the need for task support actions, especially information display support for working memory, and produced computer process specifications for human task support. Dialogue specification for a direct manipulation interface design was taken from the JSD object/event model from which permissible manipulations were derived. Further PSD diagrams were constructed to specify interface object management processes. Preliminary evaluation of the method showed that the method was easy to learn even for non HCI specialists.
Keywords: Interface design, Systems design, Task analysis, Direct manipulation
Humans, Computers, and Contracts BIBAK 161-175
  A. J. Gundry
Large interactive systems are increasingly purchased by means of competitive, fixed price contracts. Under a common form of this arrangement, a purchasing authority sponsors a requirement study and places a contract for a design study with two or more contractors. During the design study, the contractors are in a cost-effectiveness competition, and the one who wins will be held to his bid price for the implementation that follows. This paper looks at both sides of the contractual divide to see how HCI practice fares under these conditions. On the purchasing authority's side, the consequences of expressing HCI requirements in contractually-robust language are examined, with illustrative examples. On the contractor's side, typical constraints on his HCI design team are restricted access to users and the pressure to show that solutions are cost-effective. The paper reviews some other procedures in this context: user demonstrations, technical adjudication and acceptance tests, and outlines their implications. The paper concludes with a discussion of the challenges for HCI knowledge and practice of a contractual environment, and the comparisons to be made with other disciplines.
Keywords: HCI, Procurement, Contracts

Implementation Tools

A User Oriented Design Process for User Recovery and Command Reuse Support BIBAK 179-198
  Yiya Yang
This report discusses the typical working environment of user recovery and command reuse support and defines the range of services provided by it. A user-oriented design process for user recovery and command reuse support is described that ascribes a central role to empirical and analytical evaluation. The results of a survey of users' views upon existing and idealised user recovery and command reuse support is reported and discussed. In addition, literature informed analysis is used to explore the issues of support representation and command history organisation. Both are used to illustrate how design considerations enter into design process stages for user recovery and command reuse support. A four component architecture for such support is proposed to underpin these considerations comprising a context information base, a recovery knowledge base, an application model and a recovery manager.
Keywords: Undo, Redo, Reuse, User requirements
Issues Governing the Suitability of Programming Languages for Programming Tasks BIBAK 199-215
  Marian Petre; Russel Winder
This research was provoked by assertions in the literature about the 'obvious naturalness' of particular programming languages for general programming. It was intended to uncover principal issues governing the suitability of general purpose programming languages for expressing different types of solutions and to observe factors which obstructed coding or inhibited it altogether. The study required experts to program solutions to a variety of problems in several languages, in order to exercise their opinions and expertise. The general pattern which emerged from the protocols was that experts devised solutions not in terms of a particular programming language, but in terms of a pseudo-language which was a patchwork of different notations and approaches, implying that they found different languages appropriate for different aspects of solution, and that they used a personal computational model which was an amalgam of all their computational knowledge. Solutions so devised were coded into a given programming language, often with heavy translation overheads, particularly for data structures. Once a satisfactory algorithm was adopted, experts resisted a change of algorithm unless provoked strongly. Three sources of irritation in coding were reported consistently: inadequate data structuring tools, inefficiency, and poor interaction facilities.
Keywords: Naturalness, Coding, Computational model
SEE: A Safe Editing Environment; Human-Computer Interaction for Programmers BIBAK 217-233
  J. Rodger Harris
User-centred design of interactive systems requires many iterations of design and implementation. Software engineering methodologies for software development base their approach on the life cycle and stepwise refinement which, it is assumed, dictate an orderly development.
   One of the problems software engineers must overcome is the control and review of multiple versions which are needed for comparison purposes as the development proceeds. The programmers' support provided by system designers, even when used by experts, often leads to misuse and loss of vital information. Even sophisticated source code control systems are not used consistently when developing alternative versions, so leading to loss of working examples for demonstration to users.
   This user-centred source code control for programmers is based on stepwise refinement and allows the development of alternative prototypes to be controlled by the programmer(s) in any available language. A number of design principles have been developed and put into practice; a working version has been added to an existing programming environment and evaluated with a user population of novice, intermittent and expert programmers. The system is called the Safe Editing Environment (SEE) and is implemented in the Unix operating system.
Keywords: Programming environment, Software engineering tool kit, Rapid prototyping, User centred programmer support, Interactive design, Stepwise refinement

Systems Support

User-Driven Adaptive Behaviour, A Comparative Evaluation and an Inductive Analysis BIBAK 237-255
  A. Brooks; C. Thorburn
The comparative effectiveness of user-driven adaption has remained unevaluated until now. An experiment is reported in which two groups of subjects made use of separate text-based interfaces to an operating system environment. One group made use of a traditional interface with a help system, the other made use of a user-driven adaptive interface. The latter group of subjects could move between three different interface styles by a single function key-stroke at any request for input. Both interfaces were built using CONNECT and the experiment was carried out within a research paradigm promoted by Brooks. The group using the traditional interface was found to have requested help significantly more often than the number of times the other group pressed the function keys and four of the subjects commented negatively on the entry/exit nature of the help system. User-driven adaption was otherwise found not to enhance interaction. Interpretations are placed upon subjects' behaviour at the user-driven adaptive interface and compared with those made of the results of an earlier experiment.
Keywords: User-driven, Adaption, Evaluation, Help system, Induction
Contextual Structure Analysis of Microcomputer Manuals BIBAK 257-274
  Hiroyasu Chimura; Hiroshi Kato; Hiroyuki Mitani; Takahiro Sato
With the rapid diversification and popularization of microcomputers, the necessity for good manuals is increasing. Manuals play an important role in human computer interaction. Most manuals, however, are hard to understand. It is necessary to develop methodologies, methods and technologies for improving the quality of manuals.
   The authors considered that contextual structure plays an important part in readability. From this point of view, a method was developed for manuals contextual structure analysis and evaluation applying ISM (Interpretive Structural Modeling) method which is one means for structurally modeling a system. Use of the method helps manual developers to graphically express the whole contextual structure for manuals and to find any logical inconsistency.
   This paper describes the basic idea and the method, and then demonstrates the feasibility of using the method through actual applications.
Keywords: Human computer interaction, Microcomputer manuals, Improving quality of manuals, Contextual structure analysis, ISM (interpretive structural modeling)
Information Flow in a User Interface: The Effect of Experience and Context on the Recall of MacWrite Screens BIBAK 275-289
  J. Terry Mayes; Stephen W. Draper; Alison M. McGregor; Keith Oatley
A major theoretical and practical concern in HCI is to discover and characterise what it is that users know -- of what their expertise consists. We have tested what users remember of the detailed content of the MacWrite interface. We found that even experienced users can recall little of the menu contents, even though during use those menus are the instruments of their successful performance. It seems that the necessary information is picked up, used, and discarded; it is not learned in the sense that commands are learned. More exactly, users retain only enough information for recognition, not the much greater amount required for recall. This has implications for predicting learning times (not having to learn commands even for skilled performance should make for fast skill acquisition), and for writing documentation (no need to teach what won't be learned): thus the 'information flow' view of human action (Norman & Draper [1986]) can be used to re-interpret the findings and recommendations of the 'minimal manual' approach developed by Jack Carroll and his associates (Carroll [1984a]; Carroll [1984b]).
Keywords: HCI, Information flow, Learning, Recall, Interface design, Documentation, MacWrite, Minimal manual
Can Cognitive Complexity Theory (CTT) Produce an Adequate Measure of System Usability? BIBAK 291-307
  Christine Knowles
Superficial interface characteristics alone (e.g., mouse movements, command names, syntax) cannot adequately explain novices' learning difficulties. A source of error in user/system interaction can occur when there is a mismatch between the system and the user in terms of the way in which the domain is being represented by the system and the user's ability to carry out tasks which effect changes in the domain.
   Kieras and Polson (1985), proposed that cognitive complexity theory (CCT) could provide some quantitative measure of the usability of an interface. CCT represents job-task knowledge using production rules, which in conjunction with a task-to-device mapping structure attempts to provide a formal description of both user knowledge and device behaviour. CAD systems in the fashion industry provide an interesting opportunity to assess CCT by focusing on the highly skilled design activity of pattern cutting. This study tests the basic tenets of CCT and its ability to predict errors and learning difficulties when using CAD tools for pattern cutting, and goes on to suggest that the quality of the system's representation of the domain can, in part, determine interface complexity such that a purely quantitative measure of user-task knowledge (e.g., counting production rules) is both limited in application and inappropriate as a reliable metric for evaluating sources of complexity in an interface.
Keywords: Cognitive complexity theory (CCT), User knowledge requirements, Domain representation, CAD systems, Interface complexity
Training for Optimising Transfer between Word Processors BIBAK 309-328
  Clare Pollock
This paper describes research which aims to develop a type of training programme for users changing from one word processor to another. The training seeks to maximise the positive and minimise the negative transfer of knowledge about one system to another and thus improve the users' performance on the second system. Evidence is first presented which indicates that transfer may be a problem for such users. A model is, then, described which is used to interpret this evidence and to develop different training solutions. Three types of training which can be related to the model were tested and all were found to reduce the problem. On the basis of these results, further training programmes were developed which were more efficient. The results of an experiment which compared the second set of programmes, are next described. The experiment showed that one type of training was superior to the others. This training gave the subjects low level information about the second system as well as relating it to the first. However, this effect was not consistent over all of the tested functions. The differences between the functions are interpreted in terms of the model and the utility of the model in aiding the development of training is discussed.
Keywords: Transfer, Training, Word processors, Frames
Measuring User Satisfaction BIBAK 329-338
  Jurek Kirakowski; Mary Corbett
The Computer User Satisfaction Inventory (CUSI) is a system independent evaluation metric questionnaire. It provides an indication of the individual's feelings of satisfaction along two dimensions: competence and affect. This paper presents data on the relationship between CUSI scores and other, more system dependent, metrics and discusses the role of user satisfaction in the development of user ability.
   The relationship between the CUSI profiles and other metrics indicates that CUSI measures aspects of users in a way that is neither context sensitive nor labour intensive, unlike other measures derived from, for example, console logs, interviews, and diaries.
   CUSI's two subscales of affect and competence work in accordance with what is hypothesised on the basis of the self efficacy theory of Bandura. When we look at a longitudinal profile of user adaptation to a computer system we find an initial period of rapid development characterised by increase in satisfaction ratings, followed by a relative plateau during which feelings of competence lag behind those of affect. After this plateau stage users begin to try experimenting with more advanced features of the interface.
Keywords: Evaluation, Metric, User development

Input/Output

A Review of Human Performance and Preferences with Different Input Devices to Computer Systems BIBAK 341-362
  N. P. Milner
A large number of studies exist which compare different computer input devices. Under experimental conditions no single device has been found to be consistently more appropriate than any other for Human-Computer interaction.
   An extensive literature review has been undertaken of papers which compare the performance of different input devices. In the studies reviewed, all the devices have been compared on either speed, accuracy or subjective preference or a combination of these three measures. Whilst it is accepted that there are studies which contradict one another, the following general conclusions can be drawn.
  • 1. For fixed choice, low resolution applications the most direct input device
        (e.g., a touch sensitive screen) is quickest and most liked by subjects.
  • 2. For quick and accurate selection or manipulation of high resolution objects
        indirect input devices are better than direct devices.
  • 3. There is no clear evidence to support the mouse, joystick or trackball as
        being the best high resolution indirect input device.
  • 4. In comparative studies, cursor keys and function keys perform poorly against
        other input devices.
  • 5. Experimental tasks and the specific design of the input device have a large
        effect on the empirical results.
    Keywords: Input devices, Speed, Accuracy, Subjective preference, Literature review
  • A Gesture Based Text Editor BIBAK 363-371
      L. K. Welbourn; R. J. Whitrow
    This paper describes a text editor which has been designed to mimic the usual pen and paper type of editing. Hand-drawn gestures are used to specify the editing task. The use of gestures as an interface becomes more important with the advent of the electronic paper. The user specifies the editing task and its range by drawing the gesture on a tablet. The 'ink' of the pen appears on a screen, allowing the user to see what is drawn. The recognition of the gesture is on-line in its nature and the results of the edit are displayed immediately. Typically, two horizontal lines drawn through a word will be recognised as a delete operation, and the word will be removed from the display. The editing operations described include deletion, insertion, new paragraph, block moves and page formatting. The editor has been designed to work on both cursive and ascii text. The output document from the editor is directed to a character recogniser for recognition purposes. The choice of gestures and their recognition is described and discussion of the user acceptability given.
    Keywords: Gestures, Text editing, Electronic paper
    Towards the Construction of a Maximally-Contrasting Set of Colours BIBAK 373-389
      Darren Van Laar; Richard Flavell
    Two experiments are reported. The first investigates the relationship between hue, lightness and saturation in determining colour contrast in displays, the second examines the effect of surrounding and adjacent colours on the perception of stimuli in colour displays. All subjects taking part in the experiments had normal colour vision. Hue difference between stimuli was found to exert the biggest single effect on colour contrast, with similar hues being discriminated significantly more slowly than different hues. Lightness difference also produced a significant effect in the same direction. Saturation effects were surprising in that more similar saturations were associated with significantly faster reaction times. In the second experiment strong brightness and hue context (induction) effects were observed but effects due to target size and saturation did not reach significance. The relevance of these findings to designers of colour displays is discussed.
    Keywords: Colour displays, Contrast effects, Information display, Psychophysics
    Gripe: A Graphical Interface to a Knowledge Based System which Reasons about Protein Topology BIBAK 391-406
      Kathryn Seifert; Christopher Rawlings
    GRIPE is an interactive graphical interface to a knowledge based system which reasons about the topological structure of proteins. The knowledge based system, TOPOL, derives symbolic, declarative representations of protein topology from the underlying three-dimensional coordinates of protein structural elements. The use of the topological representation rather than the complex three-dimensional displays provided by most molecular graphics systems is intended to make it easier for a biologist (or a computer program) to perceive certain kinds of structure and symmetry in proteins, thus easing analysis and comparison. In particular, the topological representation is useful for the detection of topological motifs, which are common folding patterns taken by the proteins. GRIPE was developed to facilitate the use of TOPOL by molecular biologists as it allows the user to construct graphical queries about the presence of linear and topological structures in selected proteins. GRIPE also provides facilities for viewing the three dimensional and topological structures of the proteins. The interface provides an easy and effective way to examine protein structure.
    Keywords: Graphics, Interface, Knowledge-based, Protein, Topology, Prolog
    Graphical Prototyping of Graphical Tools BIBAK 407-420
      David England
    This paper describes a tool set for the interactive specification and construction of graphical user interfaces. It combines a specification method, Object-ATN, with a painting tool to describe interface objects. Interfaces can then be simulated for user testing and evaluation. The tool set is part of the ECLIPSE Integrated Project Support Environment but is not limited to producing user interfaces for that environment.
    Keywords: Specification, Prototyping, Evaluation
    A Comparison of Hypertext, Scrolling and Folding as Mechanisms for Program Browsing BIBAK 421-435
      Andrew F. Monk; Paul Walsh; Alan J. Dix
    Hypertext removes some of the constraints of conventional linear text by providing mechanisms for physically realizing the conceptual links between related sections of material. This research examines the use of a hypertext browser with a literate program. A literate program has a sequential structure, in that it is divided into sections presented in a particular order, and a hierarchical structure, in that some sections 'use' other sections.
       Two experiments are described which compare the performance of users browsing the same program presented either as a linear or hypertext structure. In Experiment 1 one group used a hypertext browser the other two scrolling and folding browsers. The hypertext browser is shown to be inferior to the scrolling browser under these particular circumstances. In a second experiment two further groups of users were tested, one of which was provided with an overview of the hypertext structure. This manipulation removed the disadvantage demonstrated in Experiment 1. It is concluded that while hypertext presents many new opportunities to the interface designer, it also raises new problems. In particular, the importance of providing an overview or map of the hypertext structure is demonstrated.
    Keywords: Hypertext, Scrolling, Folding, Browsing, Literate programming
    Hypertext Tips: Experiences in Developing a Hypertext Tutorial BIBAK 437-451
      Lynda Hardman
    Hypertext is a next step in the sophistication of presenting text and graphics to users. One of the major HCI issues is how an author presents information to the reader in an easily comprehensible way. This is problematic because designing a hypertext is even more difficult than designing a good linear document, which is already difficult enough. Furthermore, at the present time, there are only a small number of good examples of hypertexts.
       This paper examines the development of a hypertext, written using the Guide hypertext system, for presenting a tutorial on the structure of the brain to physiology students. The paper describes a number of points that were raised during the authoring of the tutorial. These points are formed into guidelines which suggest how to structure a hypertext and how to make the layout of the information in a hypertext clearer.
       The Guide hypertext system allow links from graphics to text, whereas paper allows only links from text to graphics. The use of this extra dimension is discussed.
       The paper is intended to give advice on creating a hypertext for authors new to the concept.
    Keywords: Hypertext
    Optimum Display Arrangements for Presenting Visual Reminders BIBAK 453-464
      John M. Findlay; Simon P. Davies; Robert Kentridge; Anthony J. Lambert; Justine Kelly
    Developments in technology now allow designers to make use of a wide variety of layouts to present material at an interactive terminal. Our understanding of perceptual and cognitive processes shows that various tradeoffs will need consideration in evaluating such layouts (availability of material vs screen clutter; reliance on user memory vs use of reminders etc). We approach these through the framework of attentional switching.
       We shall report an experimental study which evaluates these tradeoffs in a frequently encountered text editing situation. Our editor may be set to either 'insert' or 'overtype' mode. This information can be displayed with varying prominence in a peripheral window on the screen, displayed as a change of cursor, or omitted from the screen display. We have monitored user interaction at a keystroke level during text editing sessions and show that the different forms of presentation of reminder information can result in substantial differences in performance.
    Keywords: Visual displays, Windows, Attention allocation, Peripheral vision, Reminders
    Flexible Intelligent Interactive-Video BIBAK 465-475
      T. Webb; D. G. Jameson
    University College and Middlesex School of Medicine are producing two interactive video (IV) discs for use in teaching Clinical and Surgical management. The system is designed to provide three modes of interaction and will have as its core an expert system on the domain covered by the disc. The combination of a knowledge-based system and video is very rich in information. The interface for such a system requires a format familiar enough for the user to assimilate it rapidly, but flexible enough to deal with the range of possible situations and combinations of information formats. Some lessons learnt from a previous IV project are discussed and some problems raised by the new system are aired. The authors have found like many before them, that paper-based information systems provide useful guidelines for effective presentation.
    Keywords: Interactive video, Expert system, Human-computer interface
    The Application of Cognitive Psychology to CAD BIBAK 477-488
      Andrew Dillon; Marian Sweeney
    The design of usable human-computer interfaces is one of the primary goals of the HCI specialist. To date however interest has focussed mainly on office or text based systems such as word processors or databases. Computer aided design (CAD) represents a major challenge to the human factors community to provide suitable input and expertise in an area where the users goals and requirements are cognitively distinct from more typical HCI.
       The present paper is based on psychological investigations of the engineering domain, involving an experimental comparison of designers using CAD and the more traditional drawing board. By employing protocol analytic techniques it is possible to shed light on the complex problem-solving nature of design and to demonstrate the crucial role of human factors in the development of interfaces which facilitate the designers in their task. A model of the cognition of design is proposed which indicates that available knowledge and guidelines alone are not sufficient to aid CAD developers and the distinct nature of the engineering designer's task merits specific attention.
    Keywords: Cognitive psychology, CAD, Interface design, Protocol analysis

    Dialogue Design

    How Much is Enough? A Study of User Command Repertoires BIBAK 491-507
      Paddy Anstey
    A critical examination of the VAX/VMS command repertoires of users of a university computing service has been possible following the automatic logging of all operating system commands issued by all users over a period of six months. After preliminary investigation, users selected from a variety of backgrounds and with considerable experience on the system were interviewed to probe the perceived adequacy of their repertoires for their particular tasks, and to determine factors which appear to affect command repertoire development.
       A surprisingly restricted command set was revealed amongst user communities in many disciplines, including some with a substantial tradition of computing -- and a common core of popular commands across all disciplines was readily identified. Users interviewed were generally satisfied with their command repertoires but it was clear from discussion that many users could be more effective if they had a greater grasp of the system, not least for 'housekeeping' activities. Whatever their attitude to computing as an activity, the users were all applications-driven and mostly gave the learning of the 'extras' a low priority relative to the many other demands on their time, even though the possible benefits were in some cases perceived.
       The findings from this study are given in detail, and the implications for organisational and software changes discussed.
    Keywords: User command repertoires, Computing service organisation, Software environments
    Generative Transition Networks: A New Communication Control Abstraction BIBAK 509-527
      Gilbert Cockton
    The sequences of operations which are possible in the use of an interactive system can be modelled with different formal structures. Human factors and software engineering both set requirements for the design or selection of these formal structures. This paper surveys the requirements for operation sequence specification techniques for User Interface Management Systems, dialogue specification and early evaluation. To date, most formal structures have been selected from control models developed for other aspects of system specification. These selections have failed to satisfy all requirements equally. A new formal structure, the Generative Transition Network is presented which has been designed to satisfy known requirements without bias.
    Keywords: UIMS, Dialogue specification, Generative transition networks, Operational sequences, Dialogue determination, Formal methods
    Text Processing by Speech: Dialogue Design and Usability Issues in the Provision of a System for Disabled Users BIBAK 529-544
      Jill Hewitt; Stephen Furner
    Commercial speech recognition systems are available as 'add-on' units for popular office micro-computers. A typical office system has been employed to provide a 'transparent' interface to an ordinary text processing package so that it can be used by the disabled. This paper describes the prototyping carried out to develop the dialogue offered by the system as a result of addressing its user performance characteristics.
    Keywords: Dialogue design, Usability engineering, Text processing, Speech recognition

    Artificial Intelligence Issues

    User Requirements for Expert System Explanation: What, Why and When? BIBAK 547-564
      Yvonne Rogers
    It is generally assumed that one of the important features of an expert system is that it provides relevant and informative explanations regarding different aspects of the system's reasoning. As yet, however, most current systems provide very poor explanation facilities. This paper reports on a study that investigated the extent and types of explanation required by novices to satisfy their needs in understanding deductions made by an expert system. Using the 'Wizard of Oz' technique where, unknown to the subject, a person provides a simulation of the system as an expert an experiment was carried out which looked at the usefulness of various types of explanation. Two types of explanation and their combination were compared. These were 1) rule-based 2) condition-based and 3) rule and condition. The results showed that all users accessed the explanation facility and that the level of user satisfaction was found to depend on the type of explanation provided. In general, the rule and condition group found the explanations to be the most satisfying and useful. A further experiment was carried out to evaluate the type of questions users ask when the dialogue was not initiated by the system. The findings from both studies are discussed in relation to the task demands and the level of user understanding.
    Keywords: Expert systems, Explanation
    Knowledge Elicitation: Dissociating Conscious Reflections from Automatic Processes BIBAK 565-579
      R. J. Stevenson; K. I. Manktelow; M. J. Howard
    One major difficulty with standard techniques of knowledge elicitation is that they require an expert to give a verbal report of his or her knowledge. This assumes that people have conscious access to all of their cognitive processes. 'Thinking aloud' techniques explicitly make this assumption. However, recent developments in cognitive psychology suggest that this assumption is not appropriate. When solving a problem, people may use fast, automatic processes which are not available to conscious reflection. These fast automatic processes can be distinguished from conscious reflection.
       Two computer programming experts were videoed while each taught a class of students some basic programming concepts. Four 'novice' students from each class were also videoed while explaining the taught material to another person. These videos recorded the fast, automatic actions of people describing concepts. A week later, each person's video was played back to them and key questions were asked about the subject's intentions at different points in the video. These interviews recorded the conscious evaluations and interpretations of the original performance. The reports given at these interviews were classified into production rules. A standard knowledge elicitation technique was also used on the same subjects and the same material. Subjects were presented with ten concept names from the lectures and carried out paired comparisons of the ten concept names. The outcome of the paired comparisons was then subjected to multidimensional scaling. The type and extent of the knowledge elicited in the two situations is compared for both experts and novices.
    Keywords: Knowledge elicitation, Experts, Novices, Thinking
    GOMS Meets STRIPS: The Integration of Planning with Skilled Procedure Execution in Human-Computer Interaction BIBAK 581-594
      Tony Simon; Richard M. Young
    In the context of modelling user behaviour in HCI, deliberate planning based on problem solving and the fluent execution of skilled procedures are usually treated as different kinds of behaviour and modelled by different kinds of model. In this paper we draw on previous work which argues that user modelling requires a different notion of planning from that commonly discussed in the Artificial Intelligence literature, and show that problem solving and routine cognitive skill can be regarded as opposite ends of the same continuum. A simple planner, making use of a flexible hierarchical representation for plans and operators, can provide a single mechanism able to generate behaviour spanning the entire spectrum. This integration of planning with routine cognitive skill offers a basis for unifying existing models of HCI and for extending their scope.
    Keywords: User models, Planning, Problem solving, Routine skill