Internet Learning Volume 3, Number 2, Fall 2014 | Page 81

Internet Learning
Figure 3. A Graph Database Schema for Threaded Discussion Data.
V - Graph Database Schema and
Technologies
Because network thinking is fundamental
to our approach, we will
preface our data analysis with a conceptual
overview of our graph database
schema, and a technical summary of the
graph technologies we used. We will reference
this schema in our discussion of each
research question.
A. Conceptual Overview: Graph Database
Schema
We engaged with the applied graph
science experts at the Aurelius
consulting group, creators of
the open-source TinkerPop graph computing
stack, to model the conversational data
as a network schema (a ‘directed property
graph’), build a graph database against
80
that schema, and design a domain specific
language (DSL) for traversing and interrogating
the threaded discussion graph. We
found several benefits to modeling the data
as a graph, as shown in Figure 3.
First, as a data structure, the graph
allows us to pose many questions in an exploratory
and intuitive manner. Second,
the familiar concept map construct eased
discussion and reasoning about the data
among more- and less-technical researchers.
This was particularly important given
that we expected to discover new and important
questions over the course of the
study. Finally, the graph-structured data is
easily exported in forms that can be used
with existing network visualization tools.
This allowed us to use visualization as a
first-class investigative tool over the course
of the study, as well as a post-hoc story-telling
tool.