160 likes | 264 Views
From OntoSelect to OntoSelect-SWSE. Paul Buitelaar, Andreas Harth DERI – NUIG, Galway February 2009. Outline. OntoSelect @ DFKI Recap of OntoSelect Functionality OntoSelect @ DERI SWSE: Semantic Web Search Engine Architecture OntoSelect-SWSE OntoSelect-SWSE Experiments
E N D
From OntoSelect to OntoSelect-SWSE Paul Buitelaar, Andreas Harth DERI – NUIG, Galway February 2009
Outline • OntoSelect @ DFKI • Recap of OntoSelect Functionality • OntoSelect @ DERI • SWSE: Semantic Web Search Engine Architecture • OntoSelect-SWSE • OntoSelect-SWSE Experiments • Ranked List of Ontologies with Rich Metadata
OntoSelect • Ontology Library and Ontology Search Service http://olp.dfki.de/OntoSelect • OntoSelect monitors the web for ontologies (indexing/updates) • Ontology browse and search (by keyword, topic, document) • Class, property and (multilingual) label browse and search • Ontology publishing (submit your ontology) • Statistics on • Formats • Human languages • Frequently used labels • Ontology publishing
OntoSelect Statistics - Multilinguality Distribution of languages in 136 ontologies with multilingual labels - out of 1530 ontologies currently collected (~9%)
OntoSelect Statistics - Labels Most frequently used labels (‘words’, ‘terms’) in 1530 ontologies
SWSE Architecture • Distributed shared-nothing architecture • Implementation scales to billions of triples
OntoSelect-SWSE Experiments • Experiment Onto • Seed set from Google & Yahoo (*.owl, *.daml, *.rdfs) • 27,519 data sources (ontologies) • 6.5m statements • Experiment Web • Crawling six degrees from seed URI http://www.w3.org/People/Berners-Lee/card • 100,555 data sources (instance data + ontologies) • 11.9m statements
Conclusion • SWSE framework facilitates web data experiments • Taking into account real-world instance usage improves ontology ranking • Web data is noisy (more data providers == more noise) • Ontology registry should include rating facility to “vote out” data sources that do not provide consensus view