Internet Learning Volume 3, Number 2, Fall 2014 | Page 104
Visualizing Knowledge Networks in Online Courses
C. RQ3 Example: Concept Progression and
Concept Overlap
Below, we illustrate two approaches we
used to explore topic focus over time:
1) Categorized concept progression;
and 2) Concept overlap. We revisit Renlit’s
thread from 7.3 RQ2 Example, in which Jakata
and Naya ask successive questions that
lead Renlit to delve deeply into the technical
applications of analytics in the wine industry.
The discussion prompt assigned students
to discuss media depictions of predictive
analytics, and to describe how analytics
are used or might be used in their own work
or industry. Figure 22 shows the graphbased
chronology for the Renlit thread from
7. RQ2 FINDINGS, for reference. Each response
node is numbered chronologically,
for easy comparison with the categorized
concept graph in Figure 23.
Concept Progression
Figure 23 illustrates a categorized concept
graph for a single discussion
thread, with Renlit as the lead author.
The twelve responses are arranged in a circle,
each labeled with its chronological order
in the discussion, and the author’s name,
ascending clockwise. 01 RENLIT is the first
post, and 12 RENLIT is the last. The grey
arrows describe the response tree structure,
and indicate where questions are present.
Edges are drawn between responses and the
concepts they mention. If a concept is only
mentioned in a single post, it floats to the
outside of that post. If a concept is mentioned
in multiple posts, it floats to the middle
and is sized according to the number of
posts that mention it (concept InDegree).
We will call these multiple-connected concepts
the ‘central’ concepts, and take them
as a high-level representation of discussion
content for purposes of analysis. You can
explore the concept graph diagram interactively
in Interactive 4. Select response nodes
and central concepts in succession to get an
idea of who is talking about what, and how
much.
In 01 RENLIT, Renlit opens the conversation
with a broad post covering all four
main concept categories, including some
media depictions of analytics, and a detailed
example of analytics in the wine industry.
The post is judged onTargetPost=true. After
Renlit quickly follows up with another media
example in 02 RENLIT, we are presented
with three question-and-answer pairs, as
shown in Figure 24. Renlit responds individually
to questions from Jakata (03), Naya
(06), and Loret (04).
In 7.3. RQ2 Example where we color-coded
the timeline diagram for questions,
spreadRequests, topicSpread, and
other attributes, we pieced together the influence
of Jakata and Naya’s questions on the
evolution of the thread. Now that we are able
to view the categorized concept graph of the
thread, we can see lexical clues to the content
of these questions and their responses.
For example, the digital ethnography indicates
that 03 JAKATA poses a question
about the use of indices in the wine industry.
Note that the dominant Wine concept
category (red) in Jakata’s question appears
to carry over into 05 RENLIT, where Renlit
answers Jakata’s question. We see a large
cloud of new wine-related concepts connected
to 05 RENLIT, including particular
wines, vintages, stock bottles, rainfall data,
neighborhood shops, Liv-Ex’s fine wine indices,
and Wine Spectator ratings, mixed in
with some business-related concepts such as
business decisions, investors, profit, dollars,
and retail. The post also connects to several
central concepts, including wine, wine business,
bottle, data, and retailer. Analytics and
Media concepts are absent. When we look
at the distribution of concept categories over
103