BTC 2011

Additional resources

• Ontology for classes in BTC dataset (extracted from data, zipped .nt file)
• Ontology for classes in BTC dataset (lookup on URIs, zipped .nt file)

For the approach introduced in our submission, we used a combination of these two ontologies.

Results

• Ontology coverage data
  • Frequency and coverage of properties of some common types.
    Schema of csv file: type;property;frequency; property rdfs:domain type as defined by BTC ontology (see above): y(es) or n(o)
• Rule data
  • Frequent Set Format: | set values | frequencies
    • e.g. | birthPlace, birthDate | 1045
  • Rule Format : |antecedent values  -> consequent values| frequency| confidence
    • e.g. |birthPlace -> birthDate| 1045 | 97%
  • Positive association rules between predicates of mo:MusicArtist 
    (0.2% support and 80% confidence; maximum rule size of 3)
  • Negative association rules between predicates of mo:MusicArtist (0.2% support and 80% confidence; maximum rule size of 3)
  • Negative and positive association rules between predicates of foaf:OnlineAccount 
    (0.2% support and 80% confidence; rules of all sizes)