Database Reference
In-Depth Information
Figure 3. Example of a DTD tree
provides examples of data coming from two RDF
data sources S 1 and S 2 , which conform to a same
RDFS+ schema describing the cultural application
previously mentioned.
Example 1: An example of RDF data
Source S1: Museum(r607); name(r607, Le
Louvre ); located(r607, d1e5); Address(d1e5);
town(d1e5, Paris ); contains(r607, p112);
paintingName(p112, La Joconde ”); Source
S2 : Museum(r208); name(r208, musée du
Louvre ); located(r208, l6f2); Address(l6f2);
town(l6f2, ville de Paris ); contains(r208, p222)
; paintingName(p222, “Iris “); contains(r208,
p232); paintingName(p232, Joconde ”);
We consider two kinds of axioms accounting
for the Unique Name Assumption (UNA) and
the Local Unique Name Assumption (denoted
LUNA). The UNA states that two data of the same
data source having distinct references refer to two
different real world entities (and thus cannot be
reconciled). Such an assumption is valid when a
data source is clean. The LUNA is weaker than the
UNA, and states that all the references related to
a same reference by a relation refer to real world
entities that are pairwise distinct.
Figure 3 is an example of a DTD of a source to
be integrated. It is represented by the tree T 1 . A
fragment of the XML document conformed to the
DTD tree T 1 is presented in Figure 4.
The Mappings
Mappings are computed in a semi-automatic way.
They are links between the ontology O and a DTD
tree D (elements or attributes). The format of the
mappings for the classes and the properties of O
is described just below.
When c 1 is a concept of O , the format of the
mappings may be:
The XML Sources
c
1 ↔ //e
c
1 ↔ //e/@att
The XML sources that we are interested in are
valid documents, instances of a DTD that defines
their structure. We consider DTDs without enti-
ties or notations. A DTD can be represented as
an acyclic oriented graph with one node for each
element definition. The links between two nodes
are composition links. The attributes associated
to the elements in a DTD are associated to ele-
ment nodes in the graph representing to the DTD.
Because the DTDs are acyclic, their associated
graph may be represented as a forest of trees,
whose roots correspond to entry points in the graph
(nodes without predecessors). Nodes shared in
the graph by several trees are duplicated in order
to make these trees independent of each other.
c
1 ↔ //e/[@att = 'val']/@att
When R is a relation between c 1 and c 2 of O
such that ∃ c 1 ↔ //a and c 2 ↔ //b, the format of
the mapping is:r 1 (c 1 , c 2 ) ↔ r 1 (//a, //a/ …/b)
When A is an attribute of c 1 represented in
the ontology O such that ∃ c 1 ↔ //a and b being
mapped to A in T , the format of the mapping is:
A of c 1 ↔ A(//a, //a/ …/b)
In this format, ↔ indicates a mapping link
between entities in O and entities in T defined by
their path using XPath (Berglund et al., 2007) in
the associated graph. e refers to an element in T ,
@att refers to the attribute att .
Search WWH ::




Custom Search