Semantic Knowledge Representations for Soft Data Fusion - Efficient Decision Support Systems: Practice and Challenges from Current to Future - page 18

5.3.2 Graph association
On figure 9, we can see an example of the need for the association phase. Graphs g 1 and
g 2 represent two TV program descriptions that we attempted to fuse. The result of the
fusion is given by the graph g 3, which depict a badly formed TV program. Indeed, this
fused TV program has two begin and two end dates. Furthermore, looking at these two TV
program descriptions, it is obvious that they are not compatible and should not be fused
because they describe two different TV programs. Our aim is to provide a method that will
enable discriminating the observations that can be fused from the ones that are obviously not
compatible thanks to the association phase.

Fig. 9. Incompatible graphs

Several similarity measures between general graphs and conceptual graphs have been
proposed (for instance in Sorlin et al. (2003), Gandon et al. (2008) and de Chalendar et al.
(2000)). Through our proposition, we focus on the local similarity of the different pairs of
nodes. We propose to compute the similarity between two graphs with regards to the best
matching of their nodes. Intuitively, we process the similarity of two graphs by maximizing
the similarity of their couples of concepts.

2

G → [ ] G

Definition 5.7. sim Graph : 0, 1 , where is a set of graphs defined on the same vocabulary, is

the function defined as follows:
Let G 1 and G 2 be two graphs to compare. C 1 (resp. C 2 ) is the set of concepts of G 1 (resp. G 2 ) and | | C 1

| | (resp. ) is the number of concepts in the graph G 1 (resp. G 2 ).

We rename G 1 and G 2 into G 1 and G 2 such that

C 2