CHI 97 Electronic Publications: Doctoral Consortium
Accounting for Individual Differences through GAMES: Guided Adaptive Multimedia Editing System
Bernd Gutkauf
Cooperative Computing and Communication Laboratory (C-Lab)
Siemens Nixdorf - University of Paderborn
Fürstenallee 11, D-33094 Paderborn, GERMANY
+49 5251 606103
gutkauf@c-lab.de
ABSTRACT
Multimedia communication is influenced by increasing complexity and reach of information and by a rapidly growing user population. Due to these developments average authors of electronically published media have little expert knowledge in multimedia presentations. They are also confronted with considerable individual differences of recipients in culture, social life, education, psychology and physiology. In order to compensate for these shortcomings it is necessary to integrate interpretation and interaction abilities of individual users into future presentation and editing systems. We are developing a chart editing system which generates critics by user request. These critics are based on a user model, on expert knowledge in chart editing and on the currently edited chart. The system helps the author to avoid commonly made mistakes. It empowers recipients to adjust certain parameters (e.g.: colors) to their individual abilities and needs.
Keywords
individual differences, perception, user model, visualization, multimedia, adaptive systems, intelligent systems, electronic publishing, cognitive psychology, computer.
© 1997 Copyright on this material is held by the authors.
INTRODUCTION
The fast dissemination of multimedia applications via World Wide Web (WWW) and PC's caused a large community of authors and recipients. Many authors using electronic media are no publishing experts. They know little about human perception and interaction on computer systems. This results often in poor quality presentations (cyan text on green background, too small fonts, etc.). Preliminary studies [2] showed also significant differences between recipients in their interpretation and interaction abilities. Computer systems offer the opportunity to improve this situation for authors and recipients.
Authors usually need guidance during the design task through a critiquing system [3] which provides relevant information on their current work. This 'general' knowledge does not need to be based on individual users. It can be derived from literature as [1] and [6]. Recipients on the other hand should be guided in adjusting a presentation to their individual abilities and preferences. Knowledge about individual user abilities can be derived through computerized tests and be maintained in the so-called 'user model'. Already implemented ability tests measure: color perception, color memory, color ranking, mental rotation and fine motor coordination of individual users.
Further discussion is based on a chart editing system which integrates expert knowledge for author-guidance and which adapts to recipients by gathering knowledge about individual abilities. I want to emphasize that similar issues are valid for other media and the interplay between those.
SCENARIO
The author starts the chart editor and begins with the work. In the middle of the work s/he has to set the colors. S/he starts the critiquing system and asks for advise. When finished s/he checks the work again by requesting additional critics. The critiquing system suggests to use a larger font for the y-axes. However, s/he ignores that critic and sends the chart (data plus meta-information) to the recipient.
The recipient loads the chart and requests critics from the critiquing system. Before checking which colors are used the critiquing system realizes that there is no information about the user's color perception. It asks the user if s/he is willing to perform a color perception test. Since the user agrees the color perception test is popped onto the screen and the user enjoys performing the game-like test. The results are stored for further usage in the user-model. The test results indicate that the author used some colors which can't be discriminated well by the user. The system reports this fact to the user and suggests alternative colors. The user clicks onto the 'Accept' button and - wow - now the chart really makes sense.
SYSTEM DESIGN
A successful system must be controllable, transparent and predictable [4] and it should be enjoyable to the user [5]. Special care was taken during system design to fulfill these requirements. The system is divided into well defined components with particular functionality. The system as a whole has adaptive behavior due to the user model. However, no component changes it's appearance or behavior towards the user. Adaptability is present through changing critics. The critiquing mechanism is initiated only by the user and critics can be explained by user request. This is especially important for achieving transparency and predictability.
It is also desirable that the system is useful and usable for novice and expert users. I am not distinguishing between novice and expert users by categorizing them but through system design. The key issue is separating the editor from the critiquing component and delivering critics only at request. Each user can decide when to request critics and will then be helped depending on the current state of work. Using that approach a novice can be guided creating a chart whereas an expert might request critics only at the end of the work or not at all.
SYSTEM COMPONENTS
Future computer systems must try augmenting user's capabilities rather than replacing them. Figure 1 describes the user as the central part of the system. The other components are placed around the user:
- User Model: It contains information about interpretation- and interaction-abilities of a user. The abilities are derived through ability tests which are implemented at the motivational level of educational games [2] (e.g.: color perception test, mental rotation test). The user model contains also preferences of a user (e.g.: the user can suppress critics which are already known).
- Critiquing System: This is a knowledge based system implemented with the 'C Language Integrated Production System' (Clips). It produces critics by user request which are based on visualization knowledge (e.g.:[1][6]), the user model, and the current state of the work (e.g.: The system knows which colors are used in the chart and produces a critic if they can't be distinguished).
- Media Editor: The media editor is a tool (in our case a chart editor) or a collection of tools to create and edit a presentation.
|
| Figure 1: GAMES -
Guided
Adaptive
Multimedia
Editing
System.
The user as central system part interacts mainly with 1) the media editor* to generate a document and 2) the critiquing system** to request critics based on expert knowledge, the user model, and the working context. Whenever it is convenient the user can perform game-like tests*** (which measure interpretation and interaction abilities) to adapt the system to individual abilities.
|
RELATED WORK AND CONCLUSION
Rule based systems like Vista [6] and PRAVDAColor [1] try to guide authors by applying rules which are applicable to average users. Our approach differs from these system in several ways. Rules are not only based on expert knowledge but also on knowledge about individual users - the user model. Instead of sending a static picture to the recipient the data and the way the chart was built are transmitted. The chart is then rebuild (usually with the same system) at the recipients site. Since author and recipient have different user models the critiquing system can react differently. The critiquing system can inform the recipient that the author used visual cues (e.g.: colors) within a chart which are not appropriate for this user.
FUTURE WORK
Currently all individual components are implemented. However, they are not well connected yet resulting in considerable overhead for the user. We will minimize this overhead through direct data exchange between critiquing system and media editor. We are also working on linking the critiquing system via object linking and embedding (OLE) to commercially available chart and slide editors.
The critiquing system is producing text-based critics with a weight factor. We will develop a graphical user interface which will allow better structuring and understanding of critics. This is a prerequisite for user testing with a diverse user population. The user tests will serve to compare current systems with the new approach.
ACKNOWLEDGMENTS
I would like to thank Stefanie Thies, the co-developer of the system and the multimedia research group at C-Lab for valuable input. Special thanks go to Gitta Domik, my advisor.
REFERENCES
- Bergman L.D., Rogowitz B.E. and Treinish L.A.: A Rule-based Tool for Assisting Colormap Selection, Proceedings of IEEE Visualization '95, 1995, Atlanta, GA, pp. 118-125.
- Domik, G.O. and Gutkauf B.: User Modeling for Adaptive Visualization Systems, Proceedings of IEEE Visualization '94, 1994, Washington D.C., pp. 217-223.
- Fischer G., Mastaglio T., Reeves B. and Rieman J.: Minimalist Explanations in Knowledge-based Systems, Proceedings of the 23rd. annual international conference on systems sciences, 1990, Kailua-Kona, HI, January 2-5.
- Höök K., Karlgren J., and Wærn A.: A Glass-Box Intelligent Help Interface. Proceedings of the 1st International Workshop on Intelligence and Multimodality in Multimedia Interfaces, 1995, Edinburgh, Scotland, July.
- Rudisill M., Lewis C., Polson P.B. and McKay T.D.: Human-Computer Interface Design, 1996, Morgan Kaufmann Publishers, Inc., San Francisco, CA, p. 269 ff.
- Senay H. and Ignatius E.: A Knowledge-Based System for Visualization Design, IEEE Computer Graphics and Applications, 1994, 14, no 6, pp. 36-47.
CHI 97 Electronic Publications: Doctoral Consortium
