Figure 4-I-1: Automatic glossary tool interface
[select this link to view image without surrounding text]
We used three interfaces within a web browser in our tests:
Figure 4-I-1: Automatic glossary tool
interface
[select this link to view image without surrounding text]
The more complex interface is shown in Figure 4-I-2.
This second interface is for the user-updateable glossary tool.
It is identical to the basic interface (shown in Figure 1)
except that it includes a toolbar at the bottom of the window
(labeled 
in Figure 2) which allows its users
to modify, delete, and add new entries to the glossary.
(Note that the labels (
… ) are not part of the
interface but are used only for the description below.)
The interface for the user-updateable glossary tool (shown in Figure 2) is divided into four sections:
The main text is displayed in the largest division, which is
located in the upper right.
A list of glossary terms is provided in the left-hand division,
while glossary entries are displayed in the bottom division.
Glossary terms in the main text are underlined and clicking on
these underlined terms displays the corresponding glossary entries
in the first of the bottom divisions (the lower division will be
discussed shortly).
Glossary entries can also be accessed by clicking on the terms
listed in the left-hand division.
A toolbar below the at the bottom of the window is
available only in the user-updateable interface.
This toolbar allows users to modify or delete exiting entries
and to add new entries to the glossary.
Figure 4-I-2: User-updateable glossary
tool interface (with parts labeled)
[select this link to view image without surrounding text]
We tried to simulate a real-world condition in which people would want to understand a text with unfamiliar terms. We asked experimental participants to read online articles about diseases with the goal of understanding their severity and treatment options. Participants were told to imagine that they, or a child in their care, had recently been diagnosed with the condition as motivation to understand the text.
The choice of health conditions, and whether participants began with a glossary or not (and which glossary they used) was randomly assigned.
When the participants had completed reading the articles and answering the questions about them, they were asked to complete a questionnaire measuring their subjective appreciation of the glossary tool they were given to use. A debriefing session followed where participants were asked to express any comments about their performance during the study. The debriefing allowed participants to openly express their views on the software and describe what areas they felt required improvement. It also proved useful in the collection of subjective data from the participants.
Forty participants from various backgrounds took part in the study. All participants were familiar with the use of computers and as such, represented the target audience for the glossary tool. Self-selection bias was not expected. Therefore, the presence of any skewing due to selection bias is not expected in the results.
The articles used for the study discussed asthma, bronchitis and influenza. These are health conditions that affect a considerable part of the general population. A large number of individuals may therefore be interested in reading about these conditions. It is likely, however, that these individuals are not familiar with the specific details presented in the articles chosen. This would give the participants incentive to read the articles with greater attention to detail.
Articles were adapted from the WebMD, [2002] website. Glossary terms were chosen manually and definitions extracted algorithmically [Kaindl et al., 1999].
The experiment was a mixed design: the comparison between the glossary and non-glossary treatment was within-subjects (in which every participant acted as their own control), but the comparison between the two glossary tools was between-subjects. A total of forty participants completed the study. Every participant undertook two sessions, each with a different glossary type and document.
We recorded:
References for works cited in this text chunk appear below. References for all works cited are available in a separate chunk.
& 
.
