VP-SE Research Group (C)
A Self-Instructional Mailer
Alfred Bork*, Bertrand Ibrahim o, Birgit Laustsen o, Bernard Levrat o
* University of California at Irvine, Educational Technology Center, Information and Computer Science, Irvine, CA 92717 USA - internet:bork@ics.uci.edu
o University of Geneva, Computer Science Dept., 24, rue du General Dufour, CH-1211 Geneva 4, Switzerland - internet: ibrahim@uni2a.unige.ch
A Postscript version of this document can be found
here.
ABSTRACT
Most software vendors consider their products to be user friendly as soon as they have an on-line help facility and, possibly, a mouse interface. Even if it is a step in the right direction, this approach is still far from being satisfactory. Experience shows that usefully using common tree-structured help implies that the user already knows what he is looking for. Such help almost never makes use of the user's previous actions to adapt its help to the current user's task. Manuals too are seldom satisfactory. If they may be useful to find a specific detail, they can in no way be sufficient to learn how to use a software product.
In a project we have started, we propose a fundamentally new approach: from the
very beginning of the design of a software product, intimately intertwine
learning how to use the product and use it at the same time. This approach
implies careful design of an adaptive user interface and a detailed modelling
of the user's knowledge. The design of this project has almost been completed,
and part of it has already been implemented.
1. Introduction
The Friendly Mailer is an electronic mail system currently being designed at
the University of Geneva and the University of California, Irvine. It will be
completely self-instructional. By self-instructional we mean
that the program will contain internally everything that is needed for all
types of users to learn how to use the system. One aspect of self-instructional
software is that it adapts to the level of each user. Almost no software
available today is self-instructional. Users have manuals or extensive text-
oriented help on-line. We do not believe that either approach is
adequate except for experienced frequent users. Context-dependent help,
hypertext, and icons do not alter this situation. Even an expert who seldom
uses a given capability is often in trouble.
Most users of computer tools are neither experienced nor frequent users.
Many users are moreover poor readers, particularly when confronted
with large amounts of text to read. Hence human-computer interfaces are
often inadequate for the majority of users. The key concept in the development
of the Friendly Mailer is highly interactive self-instructional
software, combining learning and use in a single format. Although this
project is concerned with electronic mail, the concept of self-instructional
software that adapts to the skills of users has widespread consequences
for many types of widely used software. A paper available from the authors
elucidates further the concept of self-instructional software.
2. Features
The system will be a complete electronic mail system. Access to all the
features common to full electronic mail systems will be provided. According to
X.400 terminology, the software we are developing covers
most functions of a user agent plus some more general functions. The
program will conform to the latest international standards for electronic
mail, such as X.400 or X.500.
The new mailer will be convenient for a complete novice, a casual user,
or an experienced electronic mail user. We want to avoid either being too
boring for an expert user or too difficult to use by less skilled users. The
program will exhibit a highly friendly human-computer interface. The
user interface depends on the level of the user. The friendly mailer will
behave very differently for novice users and for experienced users. This
transition will be continuous.
- The system will keep records about the capability of each user of electronic mail.
- Decisions will frequently use this history information.
- Information kept will depend on the user's level of expertise
An experienced user of mail systems will be able to customize the interface, even if he or she is not familiar with the operating system. The system itself will embody the primitives to customize the user interface.
Learning will often occur while using the mail system. So learning and
use are combined, part of the same activity, not separated as they usually
are. All learning material is part of the system.
- No manuals are necessary for any user, even a complete beginner. But
manuals will also be provided, for those users preferring them. No instruction outside the system is required.
- Learning is always interactive, unlike the usual text-based help sequences present in many computer system, such as the ``man'' pages in
UNIX. Interaction implies active learners, not passive readers. The learning process will be highly conversational.
- Learning is adapted to the level of the user. So the system will be, in a
way, ``intelligent.'' Users do not learn about a facility until it is appropriate, perhaps long after they have begun to use the system.
- New facilities are introduced as users become more skillful. At any
time, the mail system knows the capabilities and preferences of each user
of the system, and makes decisions on this basis.
Users will acquire the facilities they need as they progress. No prior
knowledge of the operating system or of other programs (such as editors)
is required. The system is self-contained. So it is not necessary for users
to acquire this additional knowledge also. This feature will be particularly important for beginners, or for infrequent users. Experienced users
could choose to use other system capabilities they are comfortable with,
such as editors, but no interactive learning capabilities would be provided in this situation. The mailer will be usable in several operating system
environments. It will be independent of the operating system, from the
viewpoint of the user. Initial development is suggested for Berkeley
Unix, Digital VAX VMS, and IBM VM/CMS. Installation details will be
system-dependent. Features that differ from a system to another will be
isolated in the programs, to ease movement to new operating systems.
The program will be easy to move to other natural languages. We intend
initially to support the product in French and English, to test the problems
of language conversion. The techniques involved have already been used
successfully in our work in computer-based learning.
3. IMPLEMENTATION
Detailed design of the material for the stages of novice and intermediate
users have been mostly completed. A prototype is currently being implemented. The planned implementation will follow the strategies used for
over twenty years in developing highly interactive computer-based learning material, first at the University of California, Irvine, and now at both
Irvine and the University of Geneva (1, 2, 3). The following steps are involved.
- overall planning of the full project. The details are similar to those in
other large contemporary software engineering projects. Emphasis is on
groups of potential users, using such tactics as brainstorming.
- detailed design of each module. Again, groups of three or four work together in this activity. Many of the designers will be those who instruct
in the use of the software to be developed. Knowledge of various types
of users is important, so that the material can react to typical user problems. A style of recording the detailed design, called a script, has proved
to be very successful.
- technical implementation. In this stage the first running programs are
prepared. Both issues of screen design and programming are involved.
We prefer that the screen design be done by competent graphic artists.
For the implementation, we are using graphical specification and automatic programming techniques that we recently developed (3), based on
the scripts mentioned above.
- formative evaluation and improvement. Several stages of formative
evaluation and improvement will be an integral part of this project. We
consider it essential to use this material with all types of users, to gather
both cognitive and affective information, and to modify the materials
based on this user information. Hundreds of typical users would try the
system in the evaluative stage (mainly university students and faculty).
Two cycles of formative evaluation and improvement are planned.
4. References
- (1)
- Bork, Alfred; Weinstock, Harold; editors,
- Designing Computer Based Learning Material,
- Springer Verlag, 1986.
- (2)
- Stephen D. Franklin, Bernard C. Levrat,
- Portability of Computer-Based-Learning Materials across Machines
and across Natural Languages.
- WCCE/85, K.A. Duncan & D. Harris Ed., North Holland, 1985
- (3)
- Ibrahim, Bertrand et al.,
- Courseware
CAD.
- WCCE/90, Conference Proceedings, pp. 383-389, North-Holland, 1990
Site Hosting: Bronco