Semantic Knowledge Representations for Soft Data Fusion - Efficient Decision Support Systems: Practice and Challenges from Current to Future - page 12

On the XML initial observations, we can see the information that we are going to fuse. For
instance, the beginning time of the TV program appears inside the

<

programme

>

marker, as

“start” attribute, the title is between the

<

title

>

markers, ...

5.2 Domain modeling / Representing knowledge with CG
In this section, we propose to use the Conceptual Graphs model in order to achieve the step
preliminary to situation recognition: domain and situation modeling . We briefly describe the
model, using the formalization proposed in Chein & Mugnier (1992) and Chein & Mugnier
(2008). As said before, the conceptual graphs model was initially proposed in order to provide
a logical system able to represent and process natural language. Therefore, it is particularly
well adapted to the representation and processing of soft data.
The conceptual graphs model is essentially composed of an ontology called the vocabulary
hereafter and the graphs themselves, containing concepts and relation nodes. We detail
hereafter these general notions and their notations.

Fig. 6. Example of conceptual graph

5.2.1 Concepts, relations and vocabulary
The term “concept” is used to refer to a concept node. The concepts represent the “things”
or entities that exist. A concept is labeled with two components: the concept's type and the
individual marker. The conceptual type defines the category to which the entity belongs. For
instance, in Figure 6 the concept [Title: “TF! Jeunesse”] is an instance of the category Title.
Its concept type is “Title”. The individual marker relates a concept to a specific object of the
world. The object represented by [Title: “TF! Jeunesse”] has the name (or value) “TF! Jeunesse” .
The first order logic form of the concept is Title “ TF ! Jeunesse ) .

The individual markers may also be undefined. An undefined or generic individual marker is
either blank or noted with a star *. It represents the existential quantifier. For instance, [Title]
or [Title : *] stands for the following first order logic expression

(

)
The term “relation” is used to refer to a relation node. The relation nodes of a conceptual
graph indicate the relationships that hold between the different entities of the situation that
is represented. Each relation node is labeled with a relation type that points out the kind of
relationship that is represented.
In this work, we consider binary relations. The arcs that link relations to concepts nodes are
arrows, allowing distinguishing the source and target concept nodes.
The notion of vocabulary was defined in Chein & Mugnier (2008), as an extension of the
support introduced in Chein & Mugnier (1992), which was itself based on Sowa's semantic
network (Sowa (1984)). The concept types and the conceptual relation types, which are used
to label the concept and relation nodes are organized in hierarchies. We restrict our approach
to relation types that are unordered. Therefore, we manage only one hierarchy that contains
the concept types.
The partial order that holds among the set of conceptual types is interpreted as a relation of
specialization: t 1 ≤

∃

x , Title

(

x

t 2 means that t 1 is a specialization of t 2 , that is to say that any instance of
the class denoted by t 1 is also an instance of the class denoted by t 2 .