HCI Bibliography Home | HCI Conferences | HCII Archive | Detailed Records | RefWorks | EndNote | Hide Abstracts
HCII Tables of Contents: 89-1a89-1b89-2a89-2b91-1a91-1b91-2a91-2b93-1a93-1b93-1c

Proceedings of the Fifth International Conference on Human-Computer Interaction

Fullname:Proceedings of the Fifth International Conference on Human-Computer Interaction jointly with the Ninth Symposium on Human Interface (Japan)
Editors:Michael J. Smith; Gavriel Salvendy
Location:Orlando, Florida
Dates:1993-Aug-08 to 1993-Aug-13
Publisher:Elsevier Science
Standard No:ISBN 0-444-89540-X ISSN 0921-2647; hcibib: HCII93
  1. HCII 1993-08-08 Volume 2
    1. Plenary Paper
    2. I. Software Interfaces
    3. II. Software Tools

HCII 1993-08-08 Volume 2

Plenary Paper

Human-Robot-Computer Interaction in Multiagent Environment BIBA 2-7
  Yuichiro Anzai
Lots of interesting problems are discovered when we extend the target of research on human-computer interaction from computers, which are essentially passive, to more active machines such as autonomous mobile robots. More appear if we further extend it to communities of robots and computers that interact with humans. This paper discusses some issues on human-robot-computer interaction in multiagent environments, and describes systems designed and implemented in our project as possible solutions to some of those issues.

I. Software Interfaces

Is User Interface Design Just Common Sense? BIBA 9-14
  Thomas S. Tullis
This paper presents the results of a test to determine whether or not experienced user interface developers could pick the best user interface for a particular task just using "common sense." In an earlier study, seven different user interfaces for performing a task involving reordering fields in a table were empirically tested. The interfaces studied covered a wide spectrum of styles, including dragging and dropping, button-pressing, and data entry. In the present study, 28 experienced programmers were shown screen images of the seven user interfaces and asked to rank order them from "best" to "worst" for this particular task. Overall, there was virtually no correlation between their rankings and the actual data from the earlier study. The developers ranked the interfaces involving dragging and dropping consistently better than they actually were, while they ranked one of the data entry interfaces consistently worse than it actually was. The results indicate that, at least for this task, good user interface design is not just common sense.
The Need for New Application Specific Interface Elements BIBA 15-20
  J. Gulliksen; M. Johnson; M. Lind; E. Nygren; B. Sandblad
The design of user interfaces for skilled workers in professional work settings should be based on style-guides that certify efficiency. Most of todays style-guides and design guidelines over-emphasise general aspects or aspects relevant to novices. To increase efficiency both of the design process and of the resulting interface, more domain specific interface elements should be used. This paper explains the basic ideas of such domain specific style-guides and gives some examples from the health care domain.
Development of a Usable Graphical User Interface Design Guide BIBA 21-26
  Robert M., Jr. Schumacher; Arnold Lund
A corporate reality, that poor user interface design negatively affects employees and customers, led us to develop a graphical user interface design guide. We discovered in developing the document that current reference materials are not very helpful and are hard to use. We considered several areas for improvement, including using copious examples, providing behavioral rationale for choice of interface controls, etc. Our experiences so far have been positive and we hope will result in achieving our desired results of improving overall interface design.
A Design Guideline Search Method that Uses a Neural Network BIBA 27-32
  Kaori Ueno; Katsuhiko Ogawa
Human-computer interface design guidelines are useful for developing well designed interfaces. The method of effectively retrieving guidelines appropriate for the designers' problems by using neural networks was examined to improve the productivity of software designers. Two learning methods for neural networks are proposed and the performance of each are compared in this paper. First, from software designers, who had used the guidelines experimentally, we found the relations between the words and guidelines the designers associated with a sample HI designed display. Next, many networks were trained using the relations, and the trained networks were evaluated using a criterion of appropriateness which had been defined in advanced. As a result, we found Method D which trained networks using the relations between each word and several guidelines was more effective than Method C which trained networks using the relations between several words and one guideline.
Designing of Highly Effective "Human-Computer" Systems, Based on Multifunctional Elements BIBA 33-37
  S. Tsiramua; R. Kashmadze
A method of the increase of efficiency of a multicomputer system "Human-Computer", functioning in the mode of simultaneous performance of functions, by means of its designing on the basis of interchangeable multifunctional elements (MFE) is presented.
   Models are given of a quantitative evaluation of effectiveness indices (flexibility and structural reliability) of functioning of MFE and of the system, designed on their basis.
Dynamic Representation of Icons in Human-Computer Interaction BIBA 38-43
  Kazunari Morimoto; Takao Kurokawa; Takeshi Nishimura
The present paper reports on two experiments that examine the psychological effects of dynamic icons and propose the desirable size and speed of icon animation representation in human-computer interaction. The first experiment compared two types of icon representation: conventional static representation and dynamic representation. Comprehensibility of icons' image and function is evaluated by a rating method. The second experiment reveals what is a desirable speed and size of dynamic representation. Results show that dynamic representation has larger effects on understanding of and feeling toward icons. The comprehensibility of dynamic icons is influenced not only by the size and speed of animated images but also by the difference between drawn objects. We propose requirements in designing dynamic icons: the size should be above 36x36 pixel area and the speed should be about 10 frames per seconds on the condition that the number of cels is under ten.
A User-Oriented Test of Icons in an Educational Software Product BIBA 44-49
  S. Fullerton; A. J. Happ
Educators evaluated two sets of icons (41 total), designed for an educational software environment, for appropriateness, discrimination, and recognition and recall. The icon sets differed in the types of objects represented. Appropriateness and discrimination indicated a clear separation of the icons, which provided direction for redesign within a product development schedule. Recognition and recall provided little distinction of the icons, probably because of the limited time available in which to run the test.
Design of the User Interface of a Collaborative Text Writing System BIBA 50-55
  Jose A. Pino; Edgardo Fabres
This paper describes the design of a software system which will provide tools to produce a text document with a group of people. It will operate on a network of distributed workstations and will have some novel features in its human-computer interface.
Visualizing Search Results: User Interface Development for the Project Envision Database of Computer Science Literature BIBA 56-61
  Lucy Terry Nowell; Deborah Hix
Project Envision, a large research effort at Virginia Tech, focuses on developing a user centered, multimedia database from the computer science literature, with full text searching and full content retrieval capabilities. Available bibliographic databases and on-line public access catalogs present search results as lists of text. We describe the Envision search results display, which presents search results in a Graphic View window as a scatterplot of document icons, with the semantic value of six icon attributes under user control. Bibliographic information about user-selected documents is displayed in an Item Summary window, while the document abstract and other user-selected data are available in a Preview Item window. Results of formative usability evaluation are discussed.
Effect of Image Presentation to the Cognition of Plural Speech BIBA 62-67
  Hiroshi Tamura; Yue Chen; Yu Shibuya
The cognitive capability of the human to the plural speech words presented simultaneously is experimentally studied in this report. For two words presented one from right and the other from the left, higher recognition rate is confined for the right ear. The right/left difference in recognition rate is higher for the untrained subjects. When two speech words are mixed electronically before presentation, trained subjects responded with high recognition rate. When facial image of the talker is associated in presentation, the words associated with image showed certain increase in recognition rate.
A Pen-Based System to Input Correct Answers to Assist in the Development of Recognition and Understanding Algorithms of Ink Data BIBA 68-73
  Yasuhiro Cho; Toshihiro Morita; Seiichi Higaki; Shinji Moriya
This system assigns information in the form of correct answers to strokes of ink data, for simplifying the development of recognition and segmentation algorithms. When performing recognition and segmentation experiments using these algorithms, such correct answers (assigned by using our system) are utilized in speeding up these experiments and in enhancing their accuracy. In this paper, we first describe this system and its working, then we explain the correct answers actually assigned and the method of assigning them, and lastly we mention the experiments that we conducted assess this system.
A New Approach to Visual Programming in User Interface Design BIBAK 74-79
  Jurgen Herczeg; Hubertus Hohl; Matthias Ressel
To face prevailing problems with existing tools for interactively building graphical user interfaces we present a new object-oriented approach to implementing visual programming tools. This approach is employed by the user interface development environment XIT. It is based on the representation of knowledge for creating and manipulating interaction objects in the underlying user interface toolkit. This knowledge forms the basis for a set of higher-level tools, including interface builders, inspectors, browsers, and tracers, which may be applied to user interfaces created by either visual or conventional programming.
Keywords: Graphical user interfaces, User interface development environments, Interface builders, Toolkits, Visual programming, Direct manipulation
Visual Interference with User's Tasks on Multi-Window System BIBA 80-85
  Hirohiko Mori; Yoshio Hayashi
This paper describes the interferences of peripheral window, which are the displayed windows where users do not perform their task, with user's task performance on a window where they perform them, from the viewpoint of visual information processing, especially, of the relationship between the foveal vision and the peripheral vision. The results of our experiments indicate that the multi-window systems interfere the user's activity of his/her main task and some factors such as the number, the layout and the characteristics of each component of the windows on the display have effects on the interference. We discuss about the causes of the interference from the viewpoint of the nature of the human visual systems, especially the relationships between the foveal and peripheral vision.
Designing Interfaces of Autonomous Agents BIBA 86-91
  Takahiro Yakoh; Nobuyuki Yamasaki; Yuichiro Anzai
This paper presents the analysis of three key factors, sensitivity, models and expressiveness, for the interface design of autonomous agents. Also the paper gives qualitative evaluation of this proposal by actually implementing the three factors in a unified way on our new operating system and autonomous mobile robots. The results are satisfactory to provide an essential step towards the design principle of autonomous agents' interfaces.
A Knowledge Representation for Large Scale Integrated Circuit Failure Analysis BIBA 92-97
  Yuko Takeguchi; Takashi Torii; Shinichi Okabe
This paper describes how to represent knowledge for a large scale integrated circuit (LSI) failure analysis system, which supports analysts by guiding analysis procedures and inferring fault origins. There are three features of this representation method. The first is in dividing knowledge about failure analysis into blocks. The second is in describing the knowledge blocks using the most suitable form for the nature of each block. The third is in constructing a knowledge base by hierarchically combining the knowledge blocks. We are certain that our representation method facilitates the construction and maintenance of a knowledge base, because human experts have already accepted this method in developing a prototype system.
A Dialogue Manager as Interface between Aircraft Pilots and a Pilot Assistant System BIBA 98-103
  M. Gerlach; R. Onken
An interface for a knowledge-based support system for aircraft pilots is described which makes extensive use of speech input and speech output for communication. The system assists pilots with regard to their planning and decision tasks and provides functions for plan execution under Instrument Flight Rules.
A Taxonomy of Distortion-Oriented Techniques for Graphical Data Presentation BIBA 104-109
  Y. K. Leung; M. D. Apperley
In recent years, there has been a growing interest in the development of distortion-oriented presentation techniques for large scale information systems. Conflicting terminologies and principles in this area have given rise to confusion amongst interface designers. This paper presents a taxonomy of these graphical techniques. It highlights their differences and similarities, enabling sensible comparisons and selections to be made.
An Information-Based User Interface Architecture BIBA 110-115
  H. Diel; J. Uhl; M. Welsch
The Information-based User Interface Architecture described in this paper is an object oriented user interface architecture with support of two aspects which are insufficiently addressed in todays existing user interface architectures, namely (1) support of object collections and composite objects, and (2) a close mapping between the applications data model and the user interface.
   The Information-based User Interface Architecture implies a strong separation of application logic and user interface logic, and an abstract interface between the two.
The Impact of the Design of the Software Control Interface on User Performance BIBA 116-121
  John R. Carlson; Laura L. Hall
Modern user interfaces have made tremendous progress in improving usability. However, as the functionality provided by software systems increases in sophistication and complexity, so too do the requirements on the interface to facilitate the access of that functionality. Additionally, as the increasing scope of functionality provided by software encourages more individuals to utilize computers, so too do the requirements for the software interface to support novice users. There have been numerous factors put forward in the literature to describe the relationship between interface design and user performance. The goal of this paper is to review the relevant literature across several fields, identify the factors which contribute to the link between interface design and user performance, and integrate these factors into a conceptual model to guide further research and practice. Improving our understanding of these factors can have dramatic effects on user performance, as well as reducing the costs of the software design process.
Computer-Aided Adaptation of User Interfaces with Menus and Dialog Boxes BIBA 122-127
  U. Malinowski; T. Kuhme; H. Dieterich; M. Schneider-Hufschmidt
Contemporary interfaces include menus and dialog boxes as major interaction techniques. Particular problems with this type of interaction are involved in complex systems with a rich functionality. Inexperienced users need to learn which actions in a complex menu hierarchy relate to their tasks. For experienced users, selecting actions from menus is costly, and this effort even sums up for frequently used actions. Dialog boxes are often used to prompt for function parameters. While many functions provide a large number of adjustable parameters the majority of these are rarely changed. Even with the simplest task users have to spend a considerable amount of mental effort due to the complex structure of dialog boxes.
   Adaptive prompting addresses these problems. An adaptive action prompter helps to select the needed action by presenting the most appropriate and most likely to be chosen actions in an additional dynamic menu. Adaptive dialog boxes use highlighting and colors to guide the users' focus and by this means make setting of parameters faster and easier.
   The involved adaptation strategies are implemented considering the ideas of computer-aided adaptation. The main goal of this approach is that users are in full control of the adaptation. On the one hand, they decide which changes take place, on the other hand, they can inspect and change the underlying user and task models.
Interactive Scenarios for the Development of a User Interface Prototype BIBA 128-133
  Mona M. Kaddah
Prototyping has become quite popular as an approach to information systems development. This may be due to the advent of fourth generation software and the increase in the number of novice users finding difficulty expressing their computer requirements. While a major and key component of a prototype, is the user interface, the specification of its structure, format and dialogue style is often difficult to determine at the outset. Tools and techniques are needed to provide novice users with an awareness of alternative design scenarios and to enable their active participation in the selection of the appropriate interface. This paper addresses this need and introduces an approach based on interactive scenarios and a development tool designed as a 'front-end builder' of the interface prototype. The main objectives of this approach are to:
  • Enhance user awareness of alternate interface designs.
  • Produce a better fit between interface structure and user needs and style.
  • Speed-up interface prototyping.
  • Enhance the elicitation of user functional requirements.
  • Provide a tool that serves in tracing interface requirements for different
       problem contexts. The approach is presently being tested on two user groups in a university environment. The reaction of the users has been closely monitored illustrating to date the promising potential of this approach.
  • Intelligent User Interface for Very Large Relational Databases BIBA 134-139
      Paul E. Reimers; Soon M. Chung
    Relational databases have successfully removed the need for physical navigation. However, they have failed to provide automatic logical navigation, that is, users must specify a logical access path (also known as a join path) when formulating a query. This becomes difficult and error-prone as the size and complexity of a database structure increase, especially for the casual user who may not be familiar with the structure of the database that he is attempting to query. Also, software applications, including artificial intelligence applications, involving relational databases often require some automated means of determining join paths in order to properly construct queries. In this paper, a solution based upon the concept of maximal objects is proposed to provide the automatic logical navigation. The proposed scheme is an extension of our previous approach [3], and consists of two parts: The first part is the design of database structures for storing the metadata of the database supported by this solution. The second part is the design of a join path generator which utilizes the metadata to provide a join path for an incomplete user query. The user specifies the target tables (relations) in the "FROM" clause of an SQL query, and the join path generator returns the complete "FROM" clause and the additional "WHERE" clause statements necessary in order to properly join the target tables.
    Spoken Language Interaction: Effects of Vocabulary Size and Experience on User Efficiency and Acceptability BIBA 140-145
      Thomas W. Dillon; A. F. Norcio; Michael J. DeHaemer
    The purpose of this study is to determine the effects of vocabulary size and interface experience on the performance and acceptance of the user. Subjects performed a hands busy and eyes busy task while interacting with a speaker-dependent connected-speech recognition system with audio output. The time required to perform the task decreased significantly when the user acquired experience with the interface. A large inclusive vocabulary decreased the number of non-recognized words spoken by the user. The results of an interface-acceptance questionnaire reveal that subjects are more accepting of the spoken language interface as they gain experience.
    Screen Layout and Semantic Structure in Iconic Menu Design BIBA 146-151
      Ping Zhan; Ram R. Bishu; Michael W. Riley
    Past research on menu interfaces has been mostly concerned with textual menus. This paper is intended to address typical issues in iconic menu design, such as screen layout and semantic structure of multiple level iconic menus. Two types of iconic menu screen layouts were studied: circular and rectangular screen layouts. The semantic structures of these menus were characterized by: 2 menu sizes x 2 directions x 2 levels of depth. The design of the menus was based on existing iconic menus. In testing, subjects studied a target icon, searched for a "matching" icon by navigating through the menu structure, and finally selected the matching icon at the lowest level of the menu. Study time, response time, and error rate were used to measure user performance.
       Results indicate that screen layout and semantic structure of iconic menus significantly affect user performance, in a way that is similar to textual menus. With the same screen space, circular screen layout was superior to rectangular screen layout.
       It is suggested that iconic menu designers should probably consider to use results obtained from textual menus as references, with care taken for the differences between textual and iconic menus. Further studies need to investigate these differences before making any sound conclusions.
    Realtime Synthesis of a Realistic Anthropomorphous Agent Toward Advanced Human-Computer Interaction BIBA 152-157
      Hiroshi Dohi; Mitsuru Ishizuka
    An anthropomorphous agent is expected to play an important role toward advanced human-computer interaction. It permits wide communication spectrum between a human and a computer. We have proposed a concept called visual software agent (VSA). The VSA uses a realtime texture-mapping for interactive animated graphics. The anthropomorphous VSA with a realistic facial image naturally rocks his/her face all the time, winks his/her eyes, and talks to a user synchronized with a speech synthesizer. For reality and natural communication, it integrates several technologies such as a deformable three-dimensional wireframe model, texture-mapping, and parallel computing. This paper describes a realtime synthesis method of the realistic anthropomorphous agent using small-scale parallel transputers. The result of the prototype implementation shows that the speed of the realistic image synthesis is fast enough for the real usage.
    Iterative Prototyping of User Interfaces BIBA 158-162
      D. Felix; H. Krueger
    This paper describes an approach to design user friendly interfaces for computer based systems, especially for public use. The iterative prototyping process may not be genuinely new, however this study presents our experience with this approach. The task was to design a user-interface for a self-service terminal for train tickets, with an underlying fare system which was very complex. The system was to use a touchscreen as input media. The design process was divided into the steps problem analysis, identification of archetypes and prototyping. Prototyping was further divided into the three stages: Archetypes, screen layout and a functional model of the complete system. User tests in all prototyping phases showed that known ways of interaction are preferred, that a colourful screen is accepted better by users, and that a linear path through a program is initially better for untrained users, but is judged to be clumsy and slow for repetitive use or for experienced users. In general, the approach with small steps, involving users at several stages with tests has shown its advantages. The results of the tests are easier to interpret, as they are embedded in the whole process of development.
    Integrating Analytical and Creative Processes for User Interface Re-Design BIBA 163-168
      Yosuke Kinoe; Hirohiko Mori; Yoshio Hayashi
    This paper proposes a framework for integrating two different aspects of the redesign of artifacts, and their corresponding stages, using empirical methods: (1) the analytical process of identifying usability issues on the basis of empirical data, and (2) the creative process of formulating redesign ideas.
       An analysis-supporting tool based on the framework is also introduced. By using a formalized analysis procedure, this tool helps analysts to carry out consistent data analysis of a mass of empirical data, and to classify the usability issues in stage 1. It also provides an environment that facilitates the creative process of idea formulation in stage 2, especially, by providing an intelligent support function using the "Genetic Algorithm." By using the results of the analytical process in stage 1, this function supports the creative process whereby analysts discover latent relationships among the issues that may initially be considered irrelevant, in order to facilitate ideas for global higher-level solutions.
    Three Usability Enhancements to the Human Factors-Design Interface BIBA 169-174
      R. G. Bias; D. J. Gillan; T. S. Tullis
    In a recent paper (Gillan & Bias, 1992), two of us considered the interaction between human factors (HF) professionals and other software designers. We couched our discussion in familiar human-computer interface (HCI) terms, and then addressed the design of this human-human interface. We identified the objectives or the HF-design interface, listed requirements, and evaluated early interface designs (e.g., where HF experts are involved only at the end of the development cycle). Further, we proposed three design concepts we expected to improve the HF-design interface: education, an electronic gatekeeper, and design analysis software.
    Effects of Hypertext Topology on Navigation Performance BIBA 175-180
      Sanjay Batra; R. R. Bishu; Brian Donohue
    This is an empirical investigation of the effects of two hypertext topologies, a hierarchy and a hypertorus structure, on navigation performance under two different network entry conditions, enter at the top of the network and enter randomly at any node. It was hypothesized that hierarchy topology is better only if a user enters the network from the top and the hypertorus topology is better for random entry situations. Thirty two subjects were randomly assigned to one of the four conditions (hierarchy structure with top entry, hierarchy structure with random entry, hypertorus structure with top entry, or hypertorus structure with random entry). The subjects' task is to browse a hypertext network and retrieve facts about ten 1991 automobiles. Performance was determined by elapsed time, number of screens examined, and accuracy. Results showed a strong learning effect over the first ten trials but equivalent performance once subjects on either of the structures for all navigational performance measures except accuracy. The hypertorus topology fosters browsing more than the hierarchy topology. However, the more constrained hierarchy topology fosters more efficient navigation behavior.
    The Relative Effectiveness of Hypertext and Text BIBA 181-186
      Mark R. Lehto; Wenli Zhu; Bryan Carpenter
    User performance when using a hypertext electronic reference system was compared to that for a conventional reference book. The links in this hypertext were the same as the index entries in the corresponding book. Specific topics and particular facts were located much faster and more accurately using the hypertext system than the book, and especially so when searching for information indirectly referred to in the index. The conclusion was that hypertext appears to be superior for "reading to do" or reference type use. A second experiment compared user performance when links corresponded exactly to the original index of the book to performance when the links were generated by computer keyword searches. Strong advantages were found in speed, accuracy, and subjective ratings for links based on the author's original index.
    Implementing Adaptable Hypermedia in a Relational Database BIBA 187-189
      William Leigh; James Ragusa
    This paper reports explorations into the use of standard database technology to support applications in hypermedia and multimedia information retrieval, an application area growing in importance. Hypermedia is usually authored and browsed using proprietary software systems built for this special purpose. But there are advantages to building information retrieval systems (including hypermedia) using standardized, open database management systems technology, especially as the hypermedia base becomes large [1]. Today the "intergalactic" standard in database management is the relational model and SQL [2]. We report methods to support advanced hypermedia applications using SQL and relational database management systems.
    Interactive Haptic Interface: Two-Dimensional Form Perception for Blind Access to Computers BIBA 190-195
      S. Lee; S. F. Wiker; G. C. Vanderheiden
    Eight sighted college students tactually explored line drawings on a computer screen using an Optacon. Particular forms of primitive elements were shown to significantly affect the likelihood of correct identification of the shapes. Configuration of the tactile array does not appear to affect the perception of two-dimensional graphic forms. The findings of this study have implications in the design of tactile communication systems, especially graphic computer access systems for people with visual impairments.
    Sharing Customization in a Campus Computing Environment BIBA 196-201
      Craig E. Wills; Kirstin Cadwell; William Marrs
    This work studies the use and sharing of customization in a campus setting where a large computing culture exists, but one that has many diverse interests rather than focusing on a cohesive project. As past of this work we explore the development of a tool to facilitate sharing and make customization features more accessible and understandable to novice users.
    An Intelligent User Interface for Multimedia Mineral Retrieval System BIBA 202-207
      Dusan Cakmakov; Danco Davcev
    In this paper, an intelligent user interface for mineral multimedia database retrieval is presented. The query mechanism is based on multimedia object content search using a multimedia knowledge-based structure called cognitive network. The experimental results for the system retrieval effectiveness expressed by recall and precision parameters are also discussed.
    Selecting Colors for Dialog Boxes and Buttons in a Text Interface BIBA 208-213
      Stanley R. Page
    The purpose of this study was to identify combinations of dialog box, button, and button highlight colors that would allow users to easily identify the highlighted button in a text interface. A combination of research analysis, performance testing, and preference testing was used to provide the design data within a short time frame. A preference test showed that users preferred black text on a gray background for dialog boxes and gray text on a blue background for document windows. Research analysis was used to reduce the potential button colors to eighteen by eliminating colors that research showed could cause problems. In the button color study, 57 participants were asked to select which of two buttons was highlighted in 306 separate pairs of buttons. Dialog boxes with gray text on blue, gray on black, or gray on dark gray as the non-highlighted button colors produced the best performance scores. Preference data indicated that white on blue, white on light blue, or white on red were good choices as highlight colors.
    An Interactive Information Filter or a Trip in Hyperspace BIBA 214-219
      Constantin Thiopoulous
    The problem of "getting lost" prevents hypermedia from exploiting their full power as a flexible and user-friendly tool for the sophisticated use of large information systems. The following paper proposes a solution to this problem by presenting an interactive information filter that supports the user in tuning the granularity level of access to the stored information.
    Dialogue Design Through Modified Dataflow and Data Modelling BIBAK 220-225
      Soren Lauesen; Morten Borup Harning
    Structured methods based on dataflow diagrams and data modelling are widely used for system analysis and design, but they are not suited for dialogue design. This paper shows a method for dialogue design that is based on modified data modelling and dataflow diagrams: The usual datamodel is complemented with a forms-based or picture-based model. This model is later extended with function bubbles representing dialogue actions.
       The method allows a high degree of user participation, especially in the first steps, where the major design decisions seem to take place.
    Keywords: Dialogue design, Structured design, Dataflow, Data modelling, User participation, User datamodel, Information demands, Prototyping
    Task Analysis in Design of a Human-Computer Interface for a Ward Based System BIBA 226-230
      T. Frascina; R. A. Steele
    This paper reports on the application of Task Analysis for Knowledge Descriptions (TAKD) to an area of ward activity, namely the requesting of pathology tests, leading to a design specification for the human-computer interface for a potential ward based computer system. It is argued that TAKD has been shown to be a powerful tool for the derivation of user interface requirements in a real world situation.
    The Effect of Direction on Object-Oriented Cursor Control Actions BIBA 231-236
      Joseph D. Chase; Sherry Perdue Casali
    A number of past studies have compared user performance with a variety of cursor control devices. However, overall conclusions regarding the "best" cursor control device for a particular application are difficult to draw because the tasks used in previous comparisons have differed greedy from one another and have not necessarily included all of the factors affecting performance. One factor that has not received much attention is that of direction of cursor movement. The purpose of the portion of the present study reported herein was to determine if direction of cursor movement significantly effects target acquisition time with various input devices. A significant direction main effect as well as a number of interactions involving direction were found; with some conditions resulting in substantially degraded user performance. The results suggest that interface developers should consider the effects of direction of movement on user performance and design accordingly where rapid target acquisition is essential. These results also suggest that future empirical evaluations and comparisons of pointing devices should consider the inclusion of a direction component.

    II. Software Tools

    CoDesk -- An Interface to TheKnowledgeNet BIBAK 238-243
      Konrad Tollmar
    The Collaborative Desktop, CoDesk, consists of a set of generic tools for CSCW, Computer Supported Cooperative Work. CoDesk is an attempt to make collaboration a natural part of the daily use of a computer. Our way to achieve this is to put the user in the centre of the computing in a similar way that applications and documents are defined and visualized in the Apple Macintosh Finder metaphor of the daily-work desktop.
       TheKnowledgeNet is a vision of a system for collaboration in teams where the members have access to a common base of information, including knowledge about who-knows-what. The design of CoDesk is based on its function as an interface to TheKnowledgeNet.
       Basic principles in the CoDesk interface are object orientation, direct manipulation, a structured room metaphor, generic communication and co-editing tools.
    Keywords: Computer supported cooperative work -- CSCW, User-centered design, Distributed systems, Multimedia communication
    The Software Architecture of DIGIS BIBA 244-249
      Hans de Bruin; Peter Bouwman
    The graphical UI design environment DIGIS facilitates designers to specify all aspects of UI design, including presentation and layout constraints, dynamic behavior, and coupling the UI with the (existing) application, with direct manipulation techniques. The software architecture of the interactive systems constructed with DIGIS is based on a multi-agent interaction model, and is to a large extent automatically generated from a formal domain application model which describes the application from the UI perspective.
    Cooperative Musical Partner System: JASPER (Jam Session Partner) BIBA 250-255
      Hirokazu Kato; Sanae Wake; Seiji Inokuchi
    JASPER (Jam Session Partner), a cooperative musical partner system is proposed in this paper. JASPER performs improvisation with a human musician and enjoys its performance. The purpose of our research is to realize communication including subjective information such as emotion between human and computer. We construct a performer model in computer which simulates a human improvising process such as [ listen -> feel -> perform ].
       The performance style of JASPER is jam session. In jam session, it is very easy for performers to represent their emotion or intention and to communicate with each other through their performance. Performers represent uprush or depression of emotion in a performance. This system copes with such emotive variety as parameters named Tension Parameter.
       In experiments, we could get impressions that users could enjoy a jam session very much. Also most of them were much interested in possibilities of this system as a new style of uses of computer. We could find out the possibilities of cooperative works and communication including subjective information such as emotion between human and computer in this system.
    ESSAI -- Interactive Sales System on an IBC-Network BIBA 256-260
      D. Felix; H. Krueger
    The introduction of integrated broadband communication (IBC) to sales and information terminals offers a new range of possibilities, both to the supplier of goods or information and to the customer (the end user). The project ESSAI (Experimental Service Sale Automation on an IBC network, project R2029) which forms part of the European RACE II program is currently building such a system and is investigating the technical possibilities and the usability for the customer. A consortium of partners from Italy, France, Germany, Denmark and Switzerland is working on this project. Beside the technical realisation the main goal will be to investigate how the user can cope with the sudden access to large amounts of information and the combination of different services. The project will produce two demonstrator terminals for testing which will offer different services (tickets for cultural events such as theatre, cinema, concerts; travel tickets), goods (merchandising of products related to the services offered and other products) as well as information (how to get to places, what is on, etc.). The first will be tested this autumn in Milan (Italy), the second next year in Basel (Switzerland). New forms of help for the user will be provided (animated sequences with examples how to use the terminal, on-line voice/video transmission from a control centre of a person who will help the user). Cross-links between the different application allow users to combine the different products to a package which fulfills their precise needs.
    Design-to-Cost Support with ASCET BIBA 261-266
      J. Warschat; J. Frech
    Product costs are at least to 60%-70% [1], in some references up to 85% [2] determined by design. In fact of this, all measures for cost reduction have to join in on the design level. Because of the designer's multiple tasks, cost efficiency can only be one aspect for the designer, who has also to consider functional, material, quality or security efficiency.
       A co-ordination of these aspects on how to get low-cost, high-quality products relies on the availability of adequate information, gained from the use of one of the following categories of design analysis methods [3]:
  • Using handbooks and checklists provides rudimentary and relatively
       unstructured rules only.
  • The employment of design teams, which consist of representatives from each
       relevant area, runs all risks of time-consuming interaction problems.
  • Automatic design systems will only work on extremely narrow problems and are
       unsuitable for typical companies.
  • Design advice approaches provide information within the capabilities of a
       system, but tend to limit the designer's creativity.
  • Design rating approaches can combine the providing of relevant information
       with the full preservation of the designer's creativity. For supporting the designer in the scope of assembly and cost efficiency, the Fraunhofer Institute for Industrial Engineering has developed the software-system ASCET (ASsembly Cost Estimation Tool), which enables the designer to estimate the assembly time and costs of a product or an assembly group already during its conceiving. This, together with the estimation of material and manufacturing costs, supports the possibilities of product modification in early design stages, allowing crucial improvements by low expenditure and the dismantling of the "imaginary wall between the design and manufacturing functions". [4]
  • An Algebraic System that Symbolic Expressions Can Interact with Graphical Representations for Understanding Mathematics BIBA 267-272
      H. Saito; M. Nagata
    This paper discusses an interaction system between symbols and images for understanding mathematics. A great deal of human understanding is facilitated through the interaction between symbols and images. The system called INGRASY (INteractive system between GRAphical representations and SYmbolic expressions), is proposed as a significant contributor to the understanding of mathematics. This system is suggested to be an effective educational tool.
    Preliminary Experiment with a Distributed and Networking Card-Handling Tool Named KJ-Editor BIBA 273-278
      Naohiko Takeda; Akichika Shiomi; Kazuhisa Kawai; Hajime Ohiwa
    An experiment that four collaborators made a specification of a middle-scale software using a distributed and networking card-handling tool named KJ-Editor was conducted. The collaborators meet at a room and are provided with separate networked computers.
       According to our observation and analysis on this experiment, some features of cooperative work activity using KJ-Editor are identified; (1) a computer supported card-handling tool is a useful resource for the group in mediating their cooperative work, (2) pointing out a card or an element of the chart by a mouse has an effect for concentrating the discussion, and (3) WYSIWIS facilities sometimes become obstacles for personal viewing of the card-arrangement and cause collaborators to be uncomfortable.
    HCI Aspects of the CASSY-Environment BIBA 279-284
      Erdmuthe Meyer zu Bexten; Claudio Moraga
    In the last three years a CAD-environment for a Computer Aided symbolic Simulation SYstem, called CASSY [1], has been developed at the Fraunhofer Institute of Microelectronic Circuits and Systems in cooperation with the University of Dortmund, Chair I of the Department of Computer Science. The environment is based on the X-Window system, OSF/Motif and the UIDS Motifation [2]. This tool supports the design of signal processing systems at a high level of abstraction. The main features of CASSY are a fast, symbolic simulation, which makes possible a fast design concept verification, an interactive, ergonomical user interface and an operating mode, which is adapted to the working style of design engineers (instead of forcing engineers to adapt themselves e.g., to a rigid HDL). The implementation of consecutive prototypes of CASSY has been carried out in close contact with experienced circuit designers, who contributed to the requirements definition and to the prototypes evaluation. Moreover CASSY has proven to be able to operate with additional simulators in a multilevel fashion [3].
       Several reports on CASSY have been published in the past [4]. The reader not familiar with CASSY is kindly invited to read the proper references. In what follows, the newest developments within CASSY will be disclosed and discussed. These include the consistent duality in the representation of signals and systems, a tool to continuously update the methods library of the simulator and finally, a specification language to express both the functional and descriptive parts of a specification as well as a tool to display it and navigate through its information units.
    The TOM Approach to System Development: Methods and Tools for Task Oriented Modelling of Real-Time Safety Critical Systems BIBA 285-290
      C. P. Warren
    The Task Oriented Modelling (TOM) approach to system development is based on the production of task-related models within an overall unifying framework. The models constructed using the TOM technique represent only those aspects of the domain of application, user, and device (computer system) which are directly related to the achievement of task goals. The models are used to generate quality and cost metrics which are then used to calculate performance metrics for the task. These performance metrics direct the system developer to problem areas which can be addressed by redesign of the device, user training, or reorganisation of the domain. This performance based approach is novel in the fields of Human-Computer Interaction (HCI) and system development. As part of its support for the system development process, the TOM project has produced prototype computer-based tools which assist in the construction of the domain, user, and device models. The TOM technique was developed using the real-time safety critical domain of Air Traffic Management (ATM), the users being Air Traffic Control Officers (ATCOs). Further work is being carried out on the transfer of the TOM approach to the domain of aviation.
    Integrating CASE and UIMS for Automatic Software Construction BIBA 291-296
      Christian Martin; Christian Winterhalder
    The paper presents a tool-based architecture for the design and near-automatic construction of interactive software. AME (Application Modeling Environment) uses an object-oriented scheme for representing applications. The system integrates CASE-tool support for object-oriented analysis and design (OOAD) and a knowledge-based UIMS for automatic user interface generation and domain code integration.
    Problem Solving Support System as Thinking Acceleration Tools BIBA 297-302
      Takashi Nakamura
    In this paper, we describe some aspects of our problem solving support system. We have developed this system as thinking acceleration tools. We have developed some thinking acceleration technique, and show how a system can offer effective support through them. Especially, we propose the idea of Active Memo as mechanism for thinking acceleration. We have developed this approach through experience with a prototypical support system (the task is card game calculation).
    A Software Architecture for Cooperative Knowledge Based Systems BIBA 303-308
      A. Hadj Kacem; J.-L. Soubie; J. Frontin
    The specificity of Knowledge Based Systems (KBS) in comparison with conventional Computing Systems (CS) lies in the fact that they integrate the feature of "intelligence". Such systems incorporate the user in the problem solving and decision process [1]. As the user guides and participates in the problem solving, it becomes essential to turn the simple communication with the system into a real cooperation between the two partners. Modeling this cooperation as a communication activity at the conceptual level has been proved necessary in the development of KBS. We propose a Man-System Cooperation Model (MSCM) and a new software architecture model for Cooperative KBS. Their objective is to make easier the communication between the user and the system within the context of cooperative work. Moreover, we take into account the user's objectives and his knowledge about the domain. In this paper, we firstly present the contribution of the Seeheim model to the design of Man-Machine interfaces. Secondly, we state the specificity of KBS in relation to conventional CS. We insist, thirdly, on the interest of our MSCM. Finally, we detail the software architecture we propose for the design of cooperative KBS.
    The Implementation of Knowledge Structures in Cognitive Simulation Environments BIBA 309-314
      D. V. Benysh; R. J. Koubek
    With recent trends in labor requirements moving from manual labor to cognitive oriented tasks, the need for understanding of the factors that effect skilled cognitive task performance has never been greater. A number of methods have been developed which attempt to identify, describe, or model these factors. Additionally, some of these models have been integrated and, using a multi-factor approach, have had reasonable success in explaining factors associated with human skilled cognitive task performance.
       This research proposes one such model combining the outstanding features of Cognitive Modeling techniques into a Knowledge Organization framework. The resulting Procedural Knowledge Structure Model (PKSM) is then evaluated to assess likely structural dimensions which have an effect on task performance. These dimensions are then defined in terms of quantifiable measures, which are then empirically validated. Results indicate that the PKSM measures, and thus model dimensions, are highly significant indicators of aspects of task performance. Furthermore, these measures provide greater predictive power than traditional knowledge structure dimensions, and a combined model (with both sets of measures included) provides yet even stronger predictions. The demonstrated capability of the PKSM indicates that it potentially has implications for user modeling in the areas of Human-Computer Interaction, task design, personnel selection and training, and task analysis. Furthermore, the model may also be applicable as a design tool in Knowledge-Based Systems research.
    Experimental Method for Construction of a Knowledge-Based System for Shipping Berth Scheduling BIBA 315-320
      Masaichiro Ogawa; Norio Saito; Tsutomu Tabe; Shinji Sugimura
    This paper gives preliminary information on the methodology needed for the construction of a knowledge-based system for shipping berth scheduling. For this purpose, a computer simulator for shipping berth scheduling was developed for eliciting knowledge from humans. The results of scheduling by humans were analyzed by the protocol analysis using a GOMS model which is a useful tool to elicit knowledge for shipping berth scheduling in a real-time interactive environment. Furthermore, the elicited knowledge was transferred into a computer as a knowledge-based system for shipping berth scheduling.
    Development of Knowledge Eliciting Techniques for Expert Systems BIBA 321-324
      Shigenobu Nomura; Yoshinori Huruichi; Tatsuo Suzuki; Takashi Kondoh
    The major focus of this research is the acquisition of correct knowledge about how a skilled person makes plans in the development of the expert system. A second issue is to obtain specialized knowledge efficiently while reducing the stress on the experts who supply this knowledge.
    Structured Notations for Human Factors Specification of Interactive Systems BIBA 325-331
      K. Y. Lim; J. B. Long
    The paper identifies and illustrates the use of structured notations to support a more precise human factors specification of a system design. Structured notations are considered more suitable than formal or algebraic notations, since its graphical representations facilitates communication with users. Thus, user feedback and validation of a design may be supported better throughout system development. It is expected that the structured notations illustrated in the paper, could be used more widely since they have now been incorporated into a structured human factors method [10, 11, 12]. Off-the-shelf computer support for the notation is also available, e.g. PDF [13].
    The System is the Expert: Architecture for a Model-Based Tutor BIBA 332-337
      Michael Pearce
    This paper presents a methodology for modeling the knowledge requirements for training simulations for complex systems. By explicitly modeling the function, behavior, and connectivity of the components of a systems in one common representation, two capacities are provided. The first is a quantitative simulation of the system that the user can interact with to learn how the system behaves. The second capacity is a qualitative model of the system that be used to reason about the actions and mistakes of the user. This dual-purpose view of system models allows for a decrease in the effort required to produce a simulation-based intelligent tutoring system, compared to that associated with rule-based approaches.
    Knowledge Acquisition for a Domain-Independent Intelligent Training System BIBA 338-343
      Ronald W. Broome; Marijke F. Augusteijn
    Recent research into Intelligent Training Systems (ITS) has emphasized instruction of specific knowledge domains. An obvious next step is the development of an ITS shell. An intelligent shell should allow the incorporation of knowledge from different instructional domains into an ITS without the need for repeatedly developing intelligent features. This would enable instructional developers, with little understanding of artificial intelligence techniques, to produce an ITS solely based on their knowledge of the domain. As might be expected, there are problems associated with the design of an ITS that is not strongly coupled to a particular domain. However, progress in the development of certain domain-independent aspects [1-4] is producing several ITS shells with limited capabilities.
       ITS Challenger is a shell providing certain intelligent features for instruction in predominantly declarative domains. It employs a dual knowledge representation that allows for both an instructor's anticipation of students' needs and a domain experts point of view [4]. A set of pedagogical rules that rely upon the structure of the knowledge, but not its content, is the basis for de-coupling the system intelligence from the expert knowledge.
       A primary experimental design goal of ITS Challenger is to free the instructional developer from the need to be a programmer and a knowledge engineer. However, the knowledge representation of ITS Challenger is critical to adaptation, and a strict canonical data representation is used. Authoring tools and procedures were developed to assist an instructional designer, unfamiliar with the principles of knowledge engineering, in the effective acquisition and structuring of the knowledge needed for competent instruction by ITS Challenger.
    A Framework for Building a Knowledge Based System Using Several Experts -- With an Application for Curriculum Design of Engineering Degree Courses BIBA 344-349
      M. N. Borges; Y. Benett; M. Lewis; M. T. Thorn
    This paper recognises and justifies the need for several experts in the process of building an expert system in the context of curriculum design. The methodology is to have a Domain Expert and Subdomain Experts working independently with the knowledge engineer. The framework proposed in this paper prevents the problem of conflict of expertise by restrictions on the subdomain boundaries and limits through the concepts of input and output variables. The paper shows that this novel approach has also addressed successfully the issues of verification and validation of knowledge based systems.
    Knowledge Support Systems for Conceptual Design: The Amplification of Creativity BIBA 350-355
      Ernest Edmonds; Linda Candy
    The paper is concerned with computer-based support for conceptual design and, in particular, with the support of creative design. The nature of conceptual design is briefly reviewed and the lack of effective computer support noted. Recent developments in computer-based Knowledge Support Systems, that offer interesting possibilities, are reviewed. The study of the early design of a clearly innovative product, the Lotus bicycle, is used to inform a discussion of the requirements for Knowledge Support Systems that can support conceptual design.
    A Conceptual Model of Human Skill Requirements for Advanced Manufacturing Settings BIBA 356-361
      R. J. Koubek; G. Salvendy
    In order to achieve the economic benefits which result from immediate use of technology, personnel should be selected and trained prior to the technology implementation. This paper provides a conceptual model for determining skill requirements concurrently with the development of new technology.
    Heuristics for Generating Informative Responses to Failing User's Queries in Natural Language Database Interfaces BIBA 362-367
      Z. Trabelsi; Y. Kotani; H. Nisimura
    In this paper we describe a natural language interface to a database, incorporating a number of heuristics and a knowledge base built on top of the database, that has been developed to respond informatively to particular failing user's queries. The interface proposes a partial solution to the complex problem of handling failing user's queries. It will be shown that the interface can be ported relatively easily from one database to another. The advantages of this work over related works are also examined.
    Natural Language Interfaces for Integrated Network Management BIBA 368-372
      Raymond Chau
    This paper describes a prototype natural language query interface (NLQI) to the AtmNet Operational Support System (OSS). The AtmNet OSS is an integrated network management system for MPR's cell relay multimedia networking system for MPR's cell relay multimedia networking system based on Asynchronous Transfer Mode technology. NLQI provides a more flexible and user friendly user interface to query configuration and traffic data from the system database over the existing menu based graphical user interface (GUI). Approximately one man-year was spent to produce the prototype, which is developed using an off-the-shelf natural language application development package. The background, implementation and evaluation of NLQI are presented.
    Natural Language as Object and Medium in Computer-Based Learning BIBA 373-378
      S. Leclerc; S. de Maisonneuve
    Computer-based learning recognised the importance of user models long ago. Important works has been done on dialogue based user models and advances were made on natural language as the medium of computer-based learning. On the other hand, natural language as object of computer-based learning is a growing field. This paper intend to expose problems that typically arise when natural language is both the object and the medium of computer-based learning. Writing aids available in the market are based on relatively superficial analysis of input text and extensive elaboration of descriptive material. Our system intend to analyse in depth and elaborate a complete representation of written text with a minimal explanatory material. Finally, the system is met to reduce the explanation provided to a list of examples and contra-examples.
    Intelligent Help Facilities: Generating Natural Language Descriptions with Examples BIBA 379-384
      Vibhu O. Mittal; Cecile L. Paris
    On-line help facilities are essential in any complex system, especially for introductory or naive users. Previous studies have highlighted the need for appropriate examples along with the description. This paper describes a help/documentation facility built within an explanation framework that plans the presentation of text and examples using techniques in natural language generation. The paper shows how text and examples can influence each other and enumerates some of the other issues that arise in planning such presentations.
    Natural Language Interfaces: Specifying and Using Conceptual Constraints BIBA 385-390
      Elizabeth Godbert; Robert Pasero; Paul Sabatier
    The work described in this paper takes place in the ILLICO project which aims at the development of natural language interfaces allowing a guided composition of sentences. The time necessary for the composition of sentences is all the more reduced since the selection of the expressions is accurate at different levels: lexical, syntactic, semantic, conceptual, pragmatic. We describe here how we build and use the conceptual model of the interfaced application in order to prevent the production of conceptually incorrect sentences. In this model, we define two types of conceptual constraints: a domain constraint on a relational symbol R specifies the domains the arguments of R must belong to, and a connectivity constraint on R specifies the number of individuals that can simultaneously be interrelated by R. Respecting these constraints excludes sentences such as the dog is reading, or expressions like the dates of birth of Joan, the six grand-fathers of Peter (false existential presuppositions).
    Making Information Systems Fit Users Needs BIBA 391-396
      A. C. Gillies
    In a wide range of software technologies, authors have highlighted the need for the user perspective to be considered. However, little improvement has been noted, and even where the 'users' are themselves software developers, for example, in CASE tools, major problems are perceived. If the software community is serious about the adoption of quality management standards e.g. ISO9001, ISO9000/3, EN29001, then software must be fit for its intended purpose.
       The hypothesis underpinning this paper is that fitness for purpose is essentially an HCI issue. This paper considers existing definitions of usability and compares them with criteria suggested by a number of Information Technology (IT) practitioners in a recent study by the author. The article then goes on to consider long term fitness for purpose of software in terms of three levels of human/computer interaction.
       Finally, the article considers three different examples of information systems that have been noted for their fitness for purpose and considers the lessons that may be drawn.
    Presentation and Editing of Structured 3-D Graphics BIBA 397-402
      Steven P. Reiss
    This paper provides an overview of our efforts at providing high-quality, 3-D visualizations of information about programs. This data is generated by querying the different information sources through a single object-oriented database schema. The result of the query is a set of abstraction objects that are then mapped, through a set of user-definable translations, into a set of abstract graphical objects that represent the display. A separate package handles layout, constraints, and presentation of these graphical objects. The resultant display is interactive at three levels. Syntactic interactions allow the user to pan, zoom and fly around in 3-D space. Semantic interactions allow user actions to affect the translations from abstraction objects into abstract graphical objects and to change the set of abstraction objects by modifying the initial query.
    Rapid 3-D Editing Through Hierarchical Constraints BIBA 403-408
      John R. Rankin
    Hierarchical 3D graphical editors allow the user to create three-dimensional designs and rapidly expand on existing designs by making use of libraries of three-dimensional objects previously constructed. When the editor has the possibility of incorporating geometrical constraints into the 3D objects, development and editing is speeded up again because invalid constructions cannot be made: fixed length edges will remain of the same length, joined edges will remain joined at the same points and so on. To handle such hierarchically nested constraint systems by the traditional method of an external constraint equation solver would be increasingly difficult due to the rapidly increasing complexity and number of constraint equations to solve at one time. Additionally, this method violates the software engineering principles of information hiding and its structure does not reflect the hierarchical structure of the three-dimensional scene. An alternative approach (the "Democracy Algorithm") using distributed constraint resolution at all levels of the scene hierarchy has been found to handle increasing complexity very effectively. This paper addresses issues concerned with certain constraint-based editing of polyhedral objects.
    Reducing Repetition in Graphical Editing BIBA 409-414
      David Kurlander
    People producing illustrations with graphical editors often need to repeat the same steps over and over again. This paper describes five techniques that reduce the amount of repetition required to create graphical documents, by having the computer play a role in automating repetitive tasks. These techniques: graphical search and replace, constraint-based search and replace, constraints from multiple snapshots, editable graphical histories, and macros by demonstration, have all been implemented within the Chimera editor framework. Chimera, which contains an object-based editor for producing 2D illustrations, was built as a testbed for this research. All of these techniques are demonstrational or example-based. The user specifies concrete examples of tasks, and the system applies the tasks to other data. In addition to reducing repetition, these techniques allow users to customize the editor for the tasks that they frequently perform, and expert users to encapsulate their knowledge in a form that other users can exploit.
    Automated Construction of Application-Specific Graph Editors in an Object-Oriented Paradigm BIBA 415-420
      M. Chen; P. Townsend; C. Y. Wang
    This paper describes a development environment for constructing a class of graphical user interfaces namely application-specific graph editors. It discusses the advantages of separating the process for specifying a graph notation from that for programming a graph editor, and the importance of organising software components of graph editors in an object oriented as well as modular manner. It is concluded that the use of an object-oriented approach is the key to the automation in the process of constructing this class of graphical user interfaces.
    Using Domain Knowledge to Support Graphical Editing BIBA 421-426
      Jonas Lowgren
    The purpose of this paper is to discuss what knowledge can be used to support graphical editing and how that knowledge could be used. We present examples of how presentational, syntactic and semantic knowledge is used to support graphical editing by means of support tools in the form of critiquing systems. The paper discusses results obtained from evaluations of these support tools and indicates promising directions for future work.