Internet Learning Volume 3, Number 2, Fall 2014 | Page 90

Visualizing Knowledge Networks in Online Courses
Figure 9. Graph timeline representation of a single week’s discussion over a period of four
days, in ten-minute intervals, coded for use of citations.
That subgraph serves as the foundation
for further exploration, analysis, and visualization.
While the graph structure of
an individual author’s response corpus is
a simple, hub-and-spoke structure, the recursive,
branching structure of a discussion
thread requires a more complex traversal
that, when expanded to return all threads
associated with a collection of discussion
prompts, yields a visualization like that
shown in Figure 8.
While this early test visualization
reveals notable structural differences across
threads, and maps an intriguing geography
of citation usage, the elements of time, authorship,
and conversational content are
notably absent. We will address time and
89
authorship here, and explore content more
closely in 8. RQ3 FINDINGS.
We had neither the resources nor
the inclination to approach the problem
of time-based graph visualization programmatically
in the early phases of our
research, and existing tools were too constraining.
We therefore took an exploratory,
design-based approach to modeling conversational
graph structures in time, and
performed it on a small data set to help us
begin thinking about the problem. This visualization
approach maps conversational
terrains in a way that attempts to capture
network structure, individual response attributes,
attribute trends, and individual
participation patterns over time.