Internet Learning Volume 3, Number 2, Fall 2014 | Page 102

Visualizing Knowledge Networks in Online Courses
Figure 22. Concept Overlap Between One Post and One Assigned Text.
sponse--mentions-->concept). For some
visualizations, we also added person, resource,
and citation nodes. The addition of
resource nodes, for example, allowed us to
see the overlap between student-mentioned
concepts, and concepts in assigned reading
material, as shown in Figure 26. We applied
force-directed layouts, with concept labels
sized by the number of responses mentioning
them (concept InDegree). Based on a
close reading of the Renlit thread, we determined
that the four major concept categories
under discussion were Media, Analytics,
General Business, and Wine. We
then assigned each concept to one of those
four categories, or left it unlabeled. The categories
are color-coded, so that the resulting
visualization (Figure 23 and Interactive
4) provides a rough understanding of the
mixture of topic areas covered in each response.
Unlabeled concepts were omitted
from this visualization for simplicity.
While this approach to understanding
topical focus admittedly has its attendant
flaws and assumptions, we believe this
kind of diagram can provide some insight
into how we might gauge the prominence
of individual concepts in a conversation;
the categories of concepts under discussion;
the emergence, progression, and disappearance
of concepts over time; and the degree
to which each participant contributes to
discussion around a particular concept. The
visual elements are underpinned by real
graph relations, amenable to counting and
interpretation by algorithms. One improvement
we intend to make in ongoing work is
to relate the discussion concepts to an ontology
of the course domain, with the goal
of understanding conversational content
against the conceptual structure of course
content.
101