Research Notes
Steven Pemberton

The underlying theme to my research is "how can system architecture be
improved to the benefit of the user?" This covers a broad area, from
programming languages, to APIs, to application architectures, to document
structures.

These notes describe the work on the project "UWISH: Usability of Web-based
Information Services for Hypermedia" (http://www.cwi.nl/projects/uwish/),
which is work in combination with TNO, University of Twente - CTIT, and the
companies KPN, Océ, CAP Gemini and Rabobank.

--

Recent research by Forrester has revealed that there are 4 main reasons why
people return to websites:
1. Good content
2. Usability
3. Frequent updates
4. Fast download time

So after content, usability is the most important property of a website.

Jakob Nielsen has recently calculated that with the current increase in the
number of websites, and the existing world supply of usability experts, that
within two years each usability expert will have just one hour per website
available. On top of this, it is well known that companies producing
products typically involve usability experts too late in the process to have
a real effect. So all this means that users are going to be fairly badly
done by in the future.

The research question for Uwish then is: how can you improve the usability
of web-based services in generic ways in the light of the worsening
situation. It attempts to address it in several ways:

1. To develop cognitive engineering theory and methods identifying and
modelling navigation strategies of different classes of users, and then
using this to identify candidate generic navigation aids and to test them on
users.

2. To develop generic and specific tools, based where possible and necessary
on the techniques identified, that aid the user in using web-based services,
both actively and passively.

3. To develop formal methods of specifying user interaction with web-based
services, where possible with automatic generation of at least prototypes
from the specifications.

4. As part of knowledge transfer, to identify, collect and assemble
information on the process of producing web services with usability central.

We talk of 'web-based services', since we address more than just websites.
For instance part of our testing includes a VRML environment delivered over
the web.

Currently we have identified three navigation adaptations for further work:
'Landmarks', 'History map', and an 'Active Assistant'.

Evidence has shown that users often don't realise when they have left a
website by clicking on an external link (ask me sometime about my story with
Xaviera Hollander). Landmarks are intended to aid the user, either
positively or negatively, by indicating that the user is still in the same
area of the site, or has moved. Possible implementations are with the use of
colours, logos, or sounds.

The history in a browser is a simple stack, so that once you have backed up
in the history and branched off to another location, it is impossible to
return to places you have popped off the stack. A history map is a branching
history reflecting the structure of a site and the paths that the user has
taken through it.

Typically in attempts to be 'cool', things are often not visible to the user
on a page that would normally be helpful in navigating (like the fact that a
certain image is a link to somewhere, and what is at the end of that link).
The Active Assistant browses with you, reporting things that you cannot see
that might be helpful (using keywords you have supplied), and browsing
ahead.

We are currently building an assistant that combines all three techniques.