Internet Learning Volume 3, Number 2, Fall 2014 | Page 91
Internet Learning
Figure 10. A sample view of the discussion shown in Figure 9, with author corpora added.
First, we used Gephi to size, color,
lay out, and export the threads to SVG
format, and imported to Illustrator layers,
as described in 6. RQ1 FINDINGS. We arranged
response nodes along a horizontal
timeline of the week’s discussion based on
timestamp data, and labeled nodes with
key data points such as author name and
wordCount. We could then turn on or off
the layers containing color-coded versions
of the nodes, to reveal how the values of
automated and hand-coded response attributes
relate to the combined temporal and
graph-structural model of a conversation.
Figure 9 shows the resulting visualization
for one week of discussion. Next, we
wanted to consider each participant’s corpus
as a context for their contributions to
specific conversations. We added corpus diagrams
around the timeline, and connected
each timeline response to its position in the
author’s corpus, as shown in Figure 10.
In this visualization, each thread is
assigned a color. Each response in the thread
is circled in the same color, and a line of that
color connects the response to its position
in the author’s corpus. This allows us to see,
for example, how typical a comment is for
that author with respect to size, quality, use
of questions, depth in the discussion tree,
etc. In addition, the colored lines emanating
from a corpus diagram provide a quick
view of the extent to which that author is
participating in each of the week’s threads.
For example, Danen contributes two comments
to Danen’s own thread (orange), and
one comment each to Alakel’s, Viska’s, and
Loret’s threads (brown, green, pink). These
90