Database Reference
In-Depth Information
Table 1.
A functional summary of a selection of graph software. BG: the package
provides functionality for constructing graphs from tabular or other data (manual graph
construction excluded); DB,WS: direct access to data from databases/web services;
L&D: provides tools for the layout and display of graphs; A: provides algorithms for
the statistical analysis of graphs; Int.: interface type; BB: is web browser-based.
†
Small
graphs only.
‡
Designed for large graphs. *Limited functionality when run as an applet.
Package
BG DB WS L&D A Int. BB Summary
✓✓ ✗ ✓
†
GGobi[10]
✗ GUI ✗ General data visualisation system with
some graph capabilities
Zoomgraph[11] ✓✓ ✗ ✓
‡
✓ Text ✓* Zoomable viewer with database-driven
back end
UCINET[29]
✓
✓
✓ GUI ✗ Popular social network analysis pack-
age
✓
‡
Pajek[28]
✗
✓ GUI ✗ Analysis and visualization of large net-
works
✓
‡
Tulip[32]
✗
✓ GUI ✗ Large graph layout and visualisation
✓
‡
LGL[33]
✗
✗ GUI ✓ Large graph layout
GraphViz [34]
✗
✓
✗ Text
✗ Popular layout package
SUBDUE[14]
✗
✗
✓ Text
✗ Subgraph analysis package
structure from a set of data, without requiring SQL or other scripting com-
mands. The tool can be used to create and explore graph structures from a
variety of data sources. The graphical interface has been written as a Flash ap-
plication; the server-side code is written in ColdFusion (our primary application
development environment). The interface can also accept text-based commands
for users wishing additional flexibility.
2
Methods
The exploratory analysis process can be divided into three main stages — graph
construction; visual, interactive exploration; and the application of specific ana-
lytical algorithms. In practice, these components would be used in an interactive,
cyclical exploratory process. We discuss each of these aspects in turn.
2.1
Graph Construction
Currently, data can be accessed from one or more local or remote databases
(local in this context means “within the AADC”) or user files. Accessing mul-
tiple data sources allows a user to integrate their data with other databases,
but is predictably made dicult by heterogeneity across sources. We extract
data from local databases using SQL statements; either directly or mediated by
graphical widgets. Local files can be uploaded using http/get and are expected
to be in comma-separated text format. Users are encouraged to use standardised
column names (as defined by the AADC data dictionary), allowing the semantic
Search WWH ::
Custom Search