<is web> Information Systems & Semantic Web
University of Koblenz ▪ Landau, Germany
Semantics through Collective Intelligence
Prof. Dr. Steffen Staab
<is web>
Steffen [email protected]
Semantics though CI2 of 8
ISWeb
Collective Intelligence
Collective datasets Hosted public datasets Gated datasets
• Social networks,…
Wikipedia style Actually includes
• Discussions• Editor hierarchies• Policies
Pagerank style highly effective no coordination no control (modulo spamming)
Gene ontology DBPedia, Public census data Facebook, LinkedIn
Wikiversity FAQs
Yahoo Answers, Lycos IQ
Tagging Flickr, Delicious, … geotagging
Different Flavors
<is web>
Steffen [email protected]
Semantics though CI3 of 8
ISWeb
Collective Intelligence
Collective datasets Hosted public datasets Gated datasets
• Social networks,…
Wikipedia style Actually includes
• Discussions• Editor hierarchies• Policies
Pagerank style highly effective no coordination no control (modulo spamming)
Different Flavors
Sizes History Semantics
<is web>
Steffen [email protected]
Semantics though CI4 of 8
ISWeb
Swoogle
RDFSrulesgeo...
Billion Triples Challenge: The Power of Collective Datasets
Common approach: Import dump to new data silo
Semantic Web?
Geoquerying
GeoNames
WordNet
GeoNames
flexible
scaleable
webby
extensible
RDFS Rules
inflexiblemonolithic
notscaleable
PlaceOfBirthbirthplace
birthplace
WordNet Swoogle
fulltext
12 months in 2005/06700M triples
+ ++
+ >1Gt
<is web>
Steffen [email protected]
Semantics though CI6 of 8
ISWeb
…but not quite far enough
Stronger: Semantics is weak
) Some Collective Ontology Engineering
Bigger: There is no data like more data
) more data sources to create
) more data sources to include
Faster: Scaleability of querying
) a matter of science, not one of witchcraft!
) Impressive track record from tiny to medium size in 10 years
<is web>
Steffen [email protected]
Semantics though CI7 of 8
ISWeb
Impacts
New ways of exploring data Semaplorer (http://btc.isweb.uni-koblenz.de) Parallax (http://mqlx.com/~david/parallax/)
New ways of mining data
New ways of relating data
<is web>
Steffen [email protected]
Semantics though CI8 of 8
ISWeb
Conclusion
New forms of collective intelligence generate semantic data
) Use them!