Global Positioning System Reference
In-Depth Information
Water_system
Water_body
Spring
Lake
River_system
River
Ditch
Dry_riverbed
Perennial_river
Seasonal_river
Canal
Knar_well
Waterway
Aqueduct
Fig. 1. Example of a water system ontology.
The Wu and Palmer (1994) approach is based on the distance between
each of nodes E 1 and E 2 and their closest common ancestor E 3 as well as the
distance between E 3 and the root node as follows:
2 D node ( E 3 , root )/( D node ( E 1 , E 3 ) + D node ( E 2 , E 3 ) + 2 D node ( E 3 , root ))
where D node ( E i , E j ) for i, j = 1,2,3 is the distance, based on node counting,
between E i , E j . For instance, in Fig. 1 we consider as elements E 1 and
E 2 two nodes, River and Aqueduct , respectively. Their closest common
ancestor is the node River_system , the distance of River _ system from the
root node Water_system , i.e., D node ( River_system, Water_system ), is equal
to 3.
Similarly, D node ( River, River_system ) = 2, and D node ( Aqueduct, River_system )
= 4. Thus, the Wu and Palmer similarity measure of River and Aqueduct is
equal to 0.5.
This semantic distance has been reformulated by Resnik in (Resnik
1999) according to edge counting. The principle of this approach is based
on the distance ( D 1 and D 2 ) which separates nodes E 1 and E 2 from the root
node and the distance ( D 3 ) which separates the closest common ancestor
of E 1 and E 2 from the root node. It is given by: 2 D 3 /( D 1 + D 2 ), where D i
represents the depth of node E i (i.e., the distance of the node form the root
in a taxonomy). In the case of the example above, the Resnik similarity
measure is equal to 2
* 2
3+5 = 0.5.
The similarity measures of both Resink, Wu and Palmer approaches
score between 1 and 0. The score 1 occurs when the two concepts are the
Search WWH ::




Custom Search