244
Table 1 - Feature types of “Rio de Janeiro” instance in ADL Gazetteer Server.
Place Names
Feature Type
Rio de Janeiro, Igarape - Acre, Estado do - Brazil
streams
Rio de Janeiro - Brazil
populated
places
Janeiro, Rio
de - Brazil
streams
Rio de Janeiro - Loreto, Departamento de - Peru
populated places
Rio de Janeiro, Serra - Paraiba, Estado
da - Brazil
mountains
Rio de Janeiro, Estado do - Brazil
administrative areas
Janeiro, Rio de - Brazil
streams
Rio de Janeiro, Serra do – Brazil
mountains
SOURCE: ADL Gazetteer Server Client
As a second example, we illustrate how
to correct queries, using the
reference ontology.
Consider the following user query: “
Give me the State of Peru that has a
City called Rio de Janeiro located in”. Suppose that we use RDQL
4
(RDF
Data Query Language) as the catalog query language.
We note that one of the mayor advantages of
RDQL is its ability to mix
both metadata and instances in the same query.
Then, the above query would be formulated in RDQL as follows:
1
SELECT ?state
2 WHERE (?id,
, "Rio de Janeiro")
3 (?id, , ?state)
4 (?state, , )
5 (?state, , ?country)
6 (?country, , )
7 (?country, , "Peru")
The answer to this query will be NULL, because the first order division
of Peru is not by states, but by “Departmentos”.
This failure can be avoided using a relaxation technique (
Chu and Mao,
2000
), applied to ontologies. The relaxation technique consists in navigating
thru the class hierarchy to generalize or specialize the query terms,
expanding or reducing the scope of the user query.
The modified query will then be “ Give me the Departamento of Peru
that has a City called Rio de Janeiro located in”, which in RDQL is:
1
SELECT ?state
2 WHERE (?id, , "Rio de Janeiro")
3 (?id, , ?state)
4
RDQL:
http://jena.sourceforge.net/tutorial/RDQL/