Social Network Analysis in Community-Built Databases - Community-Built Databases: Research and Development - page 58

In different situations, we may consider a binary relation only (if object A is
related to object B or not) or we need a wider sense of relation (object A is related to
object B in degree C). This leads to unweighted (see Fig. 3.5 ) or weighted graphs
(see Fig. 3.6 ).
3.1.5 Social Networks Visualization
In the following sections, we present one particular approach for processing social
network data, which may also be suitable for processing CBD also.
As a social network we consider a set of subjects which are linked together by
some kind of relationship. Social networking - in the sense of providing services to
people to stay in touch, communicate, and express their relations - has received
great attention in recent years.
Freeman, in [ 16 ], underlines the need for social network visualization and
provides an overview of the development of their visualization. Detailed informa-
tion about the development can be found in [ 17 ]. The development from hand-
drawn images to complex computer-rendered scenes is evident, as is the shift from
classic sociograms to new approaches and methods of visualization. What remains
is the need for clarity of such visualizations.
As previously mentioned, the CBD data may be seen as two-mode networks.
Such networks can be visualized using a bipartite graph. But it is apparent from
Fig. 3.8 (which links people denoted by letters with events denoted by numbers)
that there are too many nodes and edges for this visualization method. Even if we
use a better layout algorithm (which will, e.g., place nodes in a different way to
minimize crossings between edges), the graph does not clearly illustrate the inner
structure of the data.
Paper [ 18 ] introduces the usage of the formal concept analysis (FCA), a well-
known general purpose data analysis method, in the area of social networks and
reviews the motivation for finding relations hidden in data that are not covered
by simple graph visualization. The paper shows that the Galois lattice is capable of
capturing all three scopes of a two-mode network data - the relation between
subjects, relation between events, and also the relation between subjects and events.
Fig. 3.8 Social network - visualized as bipartite graph