Controlled vocabularies build the backbone of Enterprise Semantics. Learnings from the Linked Data approach show that metadata management should be a mere decentralized work than building yet another silo which is hard to maintain. With linked enterprise vocabularies we make use of parts of the SKOS standard which foresees that thesauri and taxonomies can be linked to each other.
3. Integrate
• Integrate data more easily from different
departments/different silos
• Integrate structured data and unstructured
data
• Explore and integrate internal data with
external information
The
Collaborate
• Exchange data more easily and efficiently
challenge with external organizations – partners,
vendors, clients, etc.
Grow
• Locate and use dynamic, quickly moving
data
• Maintain a more agile knowledge base
• Expand knowledge base more easily with
the help of linked data
4. Business groups share common processes,
topics. But how do they communicate?
Research
HR
Marketing
Production
5. Is it through a common vocabulary?
Or common understanding?
Innovation
management Innovation
management
HR
Marketing
6. Globalization + Localization =
New demands on business vocabularies
Luxury car
Luxury car
Marketing
(U.S.)
Marketing
(Europe)
7. How to share common data?
What questions do business groups have in
common?
Research Production
„Innovation management methods‟ Search
Marketing/Sales HR
9. Big data linked through established
W3C semantic web standards
http://www.w3.org/standards/semanticweb/data
“Which policies in the area of renewable energy have helped to initiate projects and
programmes in the agricultural sector which finally have improved substantially the
nutritional situation in a certain country?”
11. Linked Vocabularies
• Simple Knowledge Organisation System is a W3C
standard to develop enterprise vocabularies
• SKOS provides several properties for vocabulary
linking (mapping):
– skos:exactMatch
– skos:closeMatch
– skos:broadMatch
– skos:narrowMatch
– skos:relatedMatch
http://www.w3.org/TR/2009/REC-skos-reference-20090818/
12. Example: Use of two common
phrases with similar meanings
Lean
manufacuring
Lean
production
Production
Research
13. Same concept but different contexts
http://dbpedia.org/resource/Lean_
Manufacturing
skos:exactMatch
skos:exactMatch
http://company.com http://company.com/
/research/1452 production/729
skos:exactMatch
prefLabel:Lean manufacturing prefLabel:Lean production
altLabel: Lean production altLabel: Lean enterprise
broader: Lean concepts broader: Production management
related: Scientific management, related: just in time, rationalization,
Kaizen group work
14. Linked data alignment is key
Tools for (semi-)automatic linked data alignment:
• SILK - http://lod2.eu/Project/Silk.html
• LIMES - http://lod2.eu/Project/LIMES.html
• PoolParty Thesaurus Server -
http://poolparty.biz/products/poolparty-
thesaurus-manager/
16. Querying third-party
data in your own language
„Lean manufacturing‟ Search
Research
‘Lean
enterprise’
Production
17. Querying your own data & third-party
data as if it were one single database
http://reegle.info/countries
18. Querying structured data and
unstructured data in one step
‘Show me industry news which mention countries or regions
to which our export volume has increased over the last 5 years
at least by 10% and which deal with one of our products and/or
with one of our competitors.’
(Federated) SPARQL Queries
Industry
News Export statistics
20. Explicit metadata layer
Data Data
Research Production
Metadata:
• Stored and processed separately from data
• Metadata management is part of the enterprise information management strategy
Data Data
Marketing/Sales
HR