Information Technology Reference
In-Depth Information
{
}
3. The interaction mode made the graphs
more understandable - less understandable
{
}
4. With the results of untangling the graphs, I was
very happy - very unhappy
5. When deciding which clusters to move first, the coloring made it
{
easy - hard
}
6. Coloring the graph made untangling
{
fast - slow
}
4.1
Graphs
We selected graphs that come from a real world application where graph visualization
is used for data analysis and the graph size and layoutquality requirements pose a
challenge for state-of-the-art layout methods. Our graph set consists of RNA sequence
graphs that are used for the analysis of repetitive sequences in sequencing data [11].
They have been created by running pairwise alignment on genomic sequence reads,
and to represent reads as nodes and large overlaps between reads as edges. Eight graphs
were chosen for the experiment. Participants untangled all graphs, with the conditions
randomised for each graph.
We applied the Fruchterman-Reingold algorithm
FR
to obtain the initial layouts used
in the experiments. Our goal here was to start with a layout that did not reveal the overall
graph structure completely. A completely random layoutmight pose a too difficult chal-
lenge. Using alayout generated by
FR
allows the user to identify starting points for un-
tangling while leaving enough space for improvement based on individual preferences.
Graph properties are provided in Table 1. The set of graphs can be downloaded from
http://wcl.ml.unisa.edu.au/graph-untangling/graphs.zip
.
Allgraphs have
high local density, and a sparse global structure that allows to create layouts far from hair-
balls that are showing the structure well.
Ta b l e 1 .
Overview on the graph set used for the experiment
Graph # nodes #edges density avg deg. clus. coeff. avg sh. path diameter
A
1159
6424
5.5
11.1
0.65
19.5
59
B
1748
13957
8
16
0.64
17.9
63
C
1785
20459
11.5
22.9
0.61
10.7
41
D
3010
41757
13.9
27.7
0.67
26.4
77
E
4924
52502
10.7
21.3
0.65
36
121
F
5452
118404
21.7
43.4
0.73
46.6
216
G
5953
186279
31.3
62.6
0.72
56.2
163
H
7885
427406
54.2
108.4
0.69
24.4
55
4.2
Experimental Procedure
Each participant completed the experiment in a single session. Participants were first
askedtocompleteagraph theory quiz. This quiz asked simple graph structure questions
and was designed to allow results from participants with different levels of knowledge
to be compared, and did not affect the rest of the experiment. Participants were given
instructions on how to interact with the system, and that they would have two minutes to
