Training:Advanced Ontology Engineering at FOI - 2011/Ontology Testing/Testing Solution
From Odp
(Difference between revisions)
Line 21: | Line 21: | ||
*Ontology is inconsistent because the domain of releaseDate is the intersection of Album and Track instead of the union. | *Ontology is inconsistent because the domain of releaseDate is the intersection of Album and Track instead of the union. | ||
*A track realising "some" song doesn't mean only one! | *A track realising "some" song doesn't mean only one! | ||
+ | ** We should add a restriction on the Track, under rdfs:SubClassOf: recordingOf exactly 1 | ||
*Definition of PopBand should be "equivalent class" instead of subclass restriction | *Definition of PopBand should be "equivalent class" instead of subclass restriction | ||
*The second reasoning requirement is missing, hence we cannot infer the genre of an album - can be done with an OWL 2 property chain. | *The second reasoning requirement is missing, hence we cannot infer the genre of an album - can be done with an OWL 2 property chain. | ||
*Some missing domains and ranges, missing annotations etc. | *Some missing domains and ranges, missing annotations etc. |
Revision as of 09:40, 6 October 2011
Problems in musicindustry.owl
- CQ2: The end time is missing.
- The query you would like to be able to write:
SELECT ?band ?member ?start ?end WHERE { ?x :band ?band . ?x :member ?member . ?x :membershipStartDate ?start . ?x :membershipEndDate ?end }
- But the end date property is missing, hence the only thing we can write is:
SELECT ?band ?member ?start WHERE { ?x :band ?band . ?x :member ?member . ?x :membershipStartDate ?start }
- CQ3: roles are not time-indexed, they are asserted through specific properties, e.g. bassPlayerOf
- CS4: missing name property for persons, Anthony Kiedis is using the property for bands and is inferred to be a band. Also missing the contextual statement that implies cardinality restrictions on people and bands, on the hasName property.
- Ontology is inconsistent because the domain of releaseDate is the intersection of Album and Track instead of the union.
- A track realising "some" song doesn't mean only one!
- We should add a restriction on the Track, under rdfs:SubClassOf: recordingOf exactly 1
- Definition of PopBand should be "equivalent class" instead of subclass restriction
- The second reasoning requirement is missing, hence we cannot infer the genre of an album - can be done with an OWL 2 property chain.
- Some missing domains and ranges, missing annotations etc.