Internet Learning Volume 3, Number 2, Fall 2014 | Page 92

Visualizing Knowledge Networks in Online Courses
Interactive 2. Discussion Timeline and Corpora
connections provide entry points for analysis
of the structure and evolution of conversations
across the data set. Interactive 2
enables you to browse a week of discussion
data interactively, creating custom views
like that shown in Figure 10. For example,
you can turn on and off various SKN attributes,
overlay connector lines, and explore
participation patterns, structural elements,
corpus statistics, and response typicality
with respect to author corpora.
This is a good example of a tool that
could be automated to give instructors quick
and insightful views into ongoing conversations,
enabling them to choose where and
when to interact to best effect. Such a tool
would provide an intuitive, visual way to
explore and compare response attribute distributions,
temporal patterns of interaction,
conversation structural properties, an individual’s
influence on a conversation, or the
possible impact of conversational features
or events on subsequent discourse.
The DiscussionRank Metric For Conversational
Influence
Graph traversals for reasoning about
recursively-branching response trees
will necessarily be more complex
than those we used to investigate the simple
hub-and-spoke structure of the response
corpus. One excellent example of a graph
structural property for measuring participation
and influence in a threaded discussion
is the DiscussionRank measure devised
by Marko Rodriguez of Aurelius, as part of
Aurelius’ engagement with Pearson on this
study. Roughly inspired by Google’s PageRank
algorithm, DiscussionRank measures
author influence on a discussion thread
based on a count of responses generated,
91