![](bg4.png)
boxes which accept any type of characters, including
IPA (International Phonetic Alphabet) ones. The size
of the Phraseology field in the database automatically
adjusts to the contents entered into the Phraseology
box. This allows the students to use this box freely
to enter as many collocations and colligations as they
want, but - if necessary - to also enter other pieces
of information which have no specific place in the
glossary interface, such as part-of-speech informa-
tion, links to online images and much more. Finally,
the Uses and Sources fields are dropdown menus. The
Uses menu includes three options (Technical, Popu-
lar, and Informal), and is meant to draw the student’s
attention to linguistic register; there is also the pos-
sibility to select all the three options at once. The
Sources menu invites the student to specify the pri-
mary source of information used to fill in the glossary
fields and includes the following options: Wikipedia;
encyclopaedia; monolingual dictionary; bilingual dic-
tionary; glossary; scientific/academic publication; in-
stitutional website; Linguee or Reverso; patients’
websites and blogs; other. This field may seem redun-
dant, given that an automatic tracking feature logs all
the students Web searches (Section 3.3). However, it
was considered important to make the students think
about the type of source(s) they use. Furthermore, it
may happen that a student’s searches are not tracked
because of the student working outside the proxy ser-
vice, of his refusing to be tracked, or of other tech-
nical problems. Finally, at any time the student can
delete an existing entry or term by clicking on the red
round icon in Figures 1 and 2, or he can modify an
existing entry by clicking on the pencil icon in Figure
2. Each student can create one or more glossaries, and
each glossary may include a potentially infinite num-
ber of entries, i.e. of glossary items. For this reason,
search features have also been integrated into the sys-
tem. The student can filter the glossary by topic, or
retrieve a given term by searching for key terms in the
description field or by entering the term itself. Fur-
thermore, the ‘Search all fields’ box allows students
to perform text searches in all the glossary fields, thus
making it possible, for example, to retrieve all tech-
nical terms, English terms only, or specific phrases
entered in the Phraseology field. The entire glossary
as well as the filtered results can also be downloaded
in Excel format (by clicking on the Excel icon) and
printed out, if necessary. LearnWeb glossaries can be
filled in and consulted bi-directionally (e.g. from En-
glish into Italian and viceversa). Furthermore, glos-
saries can be personal and/or collaborative and can be
shared with other students/users. In the current exper-
iment, however, the glossaries were created individu-
ally and were not shared.
3.2 Tracking Data for the Dashboard:
The Logs and the Proxy
Dumais et al. (2014) give a comprehensive overview
of behavioral log data and analysis in HCI. As they
point out, “an important characteristic of log data
is that it captures actual user behavior and not re-
called behaviors or subjective impressions of inter-
actions” (p. 350). Indirect observation methodolo-
gies like questionnaires rely on the self-evaluation of
students and may not reflect the actual work done
by the learners. On the other hand, behavioral ob-
servations are increasingly captured at a much larger
scale and can be collected in situ on a client machine
or on remote servers as people interact with appli-
cations, systems, and services. Many studies have
already used log data to analyse learning activities
(Mazza and Dimitrova, 2004; Zhang and Almeroth,
2010; Mazza et al., 2012). Most of these works utilize
only the inbuilt logging facilities of tools like Moo-
dle or WebCT. But many language learning tasks re-
quire students to search for information on websites
outside the tool used in the course. These external
actions cannot be logged by course management sys-
tems such as Moodle. Ceddia et al. (2007) used the
Weblog Analysis Tool (WAT) application to analyse
at activity level the log file data of learners’ interac-
tions collected within a Web-based learning environ-
ment (a courseware website) in order to gain infor-
mation about the tasks that the learners had engaged
in, and to determine the achievement of educational
objectives.
The study by P
´
erez-Paredes et al. (2011) goes be-
yond this and uses the log files of a proxy server
to analyze which websites students visited to ful-
fil their task. But this requires a controlled envi-
ronment, like a classroom, where the course man-
ager can fully control the Internet access or has the
privilege to install logging software on the students’
computers. Other studies have used screen-capturing
software like Camtasia or Adobe Connect to record
students’ learning activities (Bortoluzzi and Marenzi,
2017). But these recordings have to be analyzed man-
ually, thus the number of subjects is limited by the
number of evaluators. Such a method can also be con-
sidered more obtrusive than server side logging.
We have developed a novel tracking framework
that allows researchers to track all pages a student
views during a learning session without requiring
changes on the student’s computer. Our system
tracks also external websites, such as Wikipedia.org
or Google.com. Furthermore, it is not limited to class-
room use, so students can access it from home and use
online resources as they normally would. This makes
CSEDU 2018 - 10th International Conference on Computer Supported Education
148