1. WeST – Web Science & Technologies
University of Koblenz Landau, Germany
How to do things with triples?
Steffen Staab
2. What is a triple?
http://dbtune.org/musicbrainz/resource/artist/d87e52c5-bb8d-4da8-b941-9f4928627dc8
foaf:name
ABBA
WeST – Web Science & Steffen Staab
Technologies staab@uni-koblenz.de
3. What is Linked Data? The LOD Cloud
WeST – Web Science & Steffen Staab
Technologies staab@uni-koblenz.de
4. What is Linked Data? Linked Data Principles1. URIs as
identifiers
2. http
lookup
4. relations, also
to other locations
WeST – Web Science &
3. RDF
Steffen Staab
Technologies (triples)
staab@uni-koblenz.de
5. What is Linked Data good for?
Data integration is easy
Migrating different data sources to linked data is (relatively)
easy
Serendipitous use
Discover new information by following data links
Data repurposing
Querying and aggregating data can give new insights
...
WeST – Web Science & Steffen Staab
Technologies staab@uni-koblenz.de
6. Issue: From Data to Understanding
WeST – Web Science & Steffen Staab
Technologies staab@uni-koblenz.de
7. LENA – A Fresnel application
Fresnel Vocab by [Pietriga et al. ISWC-2006]
WeST – Web Science & Steffen Staab
Technologies staab@uni-koblenz.de
8. State-of-the-Art: One App at a Time
Shameless self-promotion: Semaplorer
Billion Triples Challenge 1. Prize
2008
[Schenk et al., JoWS 2009]
WeST – Web Science & Steffen Staab
Technologies staab@uni-koblenz.de
9. State-of-the-Art: One App at a Time
Shameless self-promotion: LISA
1. Prize
German
Linked Open Gov Data
Competition 2012
WeST – Web Science & Steffen Staab
Technologies staab@uni-koblenz.de
10. What‘s between the two?
One App at a Time Generic Frameworks
+ Great to use + Can be applied on all data
+ Like DB application - Data remains hard to
- Brittle understand
- Not really extensible - No process support
- Noone wants to use them
WeST – Web Science & Steffen Staab
Technologies staab@uni-koblenz.de
11. What may be a theory for doing things with triples?
PRAGMATICS
WeST – Web Science & Steffen Staab
Technologies staab@uni-koblenz.de
12. Austin: How to do things with words
Core Hypothesis by Austin:
Speech is not only passively describing a given reality, but it can
change the (social) reality it is describing through speech acts
Summary from Wikipedia, 2012-06-09
WeST – Web Science & Steffen Staab
Technologies staab@uni-koblenz.de
13. Why Do We Understand the Text Web?
WeST – Web Science & Steffen Staab
Technologies staab@uni-koblenz.de
14. Austin: How to do things with words
Phonic act / graphic act:
ThetemperatureinMilwaukeeis100°F.
Locutionary act:
temperature(Milwaukee,100)
Grice‘s
maxims
WeST – Web Science & Steffen Staab
Technologies staab@uni-koblenz.de
15. Grice‘s Maxims / Cooperative Principle
Interacting agents mutually assume that:
Quantity:
Be as informative as you possibly can,
give as much information as needed, not more.
Quality:
Criteria are
Be truthful
competing and
Pertinence:
overlapping.
be relevant,
say things pertinent to the discussion
Manner:
be clear, brief, orderly as one can
avoid obscurity and ambiguity
WeST – Web Science & Steffen Staab
Technologies staab@uni-koblenz.de
16. Austin: How to do things with words
Phonic act / graphic act:
ThetemperatureinMilwaukeeis100°F.
Locutionary act:
temperature(Milwaukee,100)
Grice‘s
maxims
Illocutionary act:
Warn the conference attendees
Perlocutionary act:
Attendees stay in the shadow, etc…
WeST – Web Science & Steffen Staab
Technologies staab@uni-koblenz.de
17. Austin: How to do things with words
Core Hypothesis by Austin:
Speech is not only passively describing a given reality, but it can
change the (social) reality it is describing through speech acts
Summary from Wikipedia, 2012-06-09
Hypothesis of this talk:
Linked data is facts, but the idea of linked data is also
re-purposing,
⟹ re-presenting,
⟹ re-narrating,
to achieve an understandable dialogue
WeST – Web Science & Steffen Staab
Technologies staab@uni-koblenz.de
18. Semantic Web / Linked Data
Phonic act / graphic act:
ThetemperatureinMilwaukeeis100°F.
- various syntaxes -
Locutionary act:
temperature(Milwaukee,100)
- RDF/OWL interpretation –
WeST – Web Science & Steffen Staab
Technologies staab@uni-koblenz.de
19. Data is not Text
Quantity:
One triple vs big data
➯ What is the right amount?
Pertinence:
Pertinence to dialogue
➯ Does the discussion/interaction determine data selection?
Manner:
Data is not sequential
➯ No implicit ordering contained in the data
(e.g. birthdata before date of death)
WeST – Web Science & Steffen Staab
Technologies staab@uni-koblenz.de
20. Semantic Web / Linked Data
Phonic act / graphic act: Lead question:
ThetemperatureinMilwaukeeis100°F. Semantic Web
Does the
- various syntaxes - have a pragmatics layer?
Locutionary act: How would this look like?
temperature(Milwaukee,100) Quantity
- RDF/OWL interpretation – Quality
Pertinence
Manner
Generic
applications easily
violate Grice‘s
maxims!
WeST – Web Science & Steffen Staab
Technologies staab@uni-koblenz.de
21. CONCEPTUAL
NAVIGATION MODEL
WeST – Web Science & Steffen Staab
Technologies staab@uni-koblenz.de
22. Hypertextual Navigation (from D. Schwabe)
What does it mean to
click here?
Semantics is clear,
but Pragmatics?
Context +
Grice‘s Pertinence!
[Bomfim & Schwabe, 2011]
WeST – Web Science & Steffen Staab
Technologies staab@uni-koblenz.de
23. Hypertextual Navigation (from D. Schwabe)
Input: LOD + Navigation Model + other stuff
Navigation Model
A Context is a set of resources that share similar
navigation opportunities.
• Context:Navigation ⇔Class:Structure+Behavior
Navigation Metamodel
WeST – Web Science & Steffen Staab
Technologies staab@uni-koblenz.de
24. Hypertextual Navigation (from D. Schwabe)
foaf:Document
AllDocuments DocumentsAlpha
DocumentsByPerson byPerson
Hitting a link of type
foaf:Person „Organization“
byDocument means different
AllPersons PersonsAlpha
things in different
contexts!
swc:AcademicEvent
Here is one!
byOrganiza on
EventsByUserProgram byUserProgram
AllEvents EventsAlpha
WeST – Web Science & Steffen Staab
Technologies staab@uni-koblenz.de
25. PROGRAMMING WITH LINKED
DATA
=
INTERACTING WITH LINKED DATA
[Schegelmann et al 2012]
WeST – Web Science & Steffen Staab
Technologies staab@uni-koblenz.de
26. Mapping Linked Data and OO Programming
Linked Data Object orientation
Meaning/Semantics Meaning/Semantics
Concept Classes
Properties Attributes/Methods
Instances Objects
Pragmatics
Classes
• Visible and used by
whom?
• Orchestration of method
calls
Responsibility-driven
WeST – Web Science & Steffen Staab
Technologies staab@uni-koblenz.de
27. Why is Linked Data Programming Tricky?
Developing Core Ontologies
Software: Core Ontology of Software (COS)
Services: Core Ontology of Software Components (COSC)
Multimedia: Core Ontology of Multimedia (COMM)
Events: Core Ontology of Events (F)
Desktop: Core Semantic Inf. Mgmt. Ontology (COSIMO)
Main Criteria:
Reusability
Plugability
Main Drawback: Programmability
WeST – Web Science & Steffen Staab
Technologies staab@uni-koblenz.de
28. What Jeopardizes Programmability?
Challenges on (RDF) knowledge representation
N-ary relations (denormalization)
Aspect on relationships („Tim has high temp, but falling“)
Roles as additional objects
(lifetime different from rigid object)
Different roles being played in different contexts
(descriptions and situations)
Solution [Schegelmann et al 2012]:
–Different ontology patterns to the rescue
–Easy to use APIs
http://www.w3.org/TR/swbp-n-aryRelations/
WeST – Web Science & Steffen Staab
Technologies staab@uni-koblenz.de
29. Pattern for Image Tagging
[Scherp&Saathoff, WWW-2010]
[Troncy et al 2007]
WeST – Web Science & Steffen Staab
Technologies staab@uni-koblenz.de
30. Ontology API Model for Image Tagging
WeST – Web Science & Steffen Staab
Technologies staab@uni-koblenz.de
32. Comparing the two structures
WeST – Web Science & Steffen Staab
Technologies staab@uni-koblenz.de
33. OntoMDE Workflow
Model of Ontologies (MoOn)
Adding declarative layer:
Structuring the ontologies into
semantic units
Ontology API Model (OAM)
Adding declarative layer:
Structuring pragmatic units specifying
how entities are to be used together
WeST – Web Science & Steffen Staab
Technologies staab@uni-koblenz.de
34. RANKING
WeST – Web Science & Steffen Staab
Technologies staab@uni-koblenz.de
35. More data here …
Which links to follow?
Where to go next?
What information is there?
What are the „good“ links …
… leading to „good“ resources?
More data here …
WeST – Web Science & Steffen Staab
Technologies staab@uni-koblenz.de
36. Relevance Ranking for Linked Data
Apply Web Ranking,
e.g. HITS, PageRank, …
Who is loved/hated most?
TripleRank
WeST – Web Science & Steffen Staab
Technologies staab@uni-koblenz.de
37. Linked Data as Tensor
Transformation
to 3-D-Tensor
Then:
[Franz et al, 2009] PARAFAC analysis
[Nickel et al, 2012]
WeST – Web Science & Steffen Staab
Technologies staab@uni-koblenz.de
38. Evaluation
16 test persons
Faceted browsing scenario
What are the most
interesting, most related,
most useful resources
(objects)?
10 queries
WeST – Web Science & Steffen Staab
Technologies staab@uni-koblenz.de
39. Evaluation: Results
1387 answers, overall inter-rater agreement: 0.7
(0 ≤ agreement ≤ 1)
8 TR
7.594
TR
7
BL
6
5
4
Doubled recall BL
3.948
TR
3
without loss of precision! 3.251
TR
2.413
BL
2
1.626 BL
1.207
1
0
Total Results Positives Negatives
WeST – Web Science & Steffen Staab
Technologies staab@uni-koblenz.de
41. Metamodels
Patterns
Pragmatics Rankings
...
Semantics
Syntax
WeST – Web Science & Steffen Staab
Technologies staab@uni-koblenz.de
42. Issue: From Data to Understanding
Cognition
Storytelling
Pragmatics
Ontology Patterns
Conceptual Modeling
Metamodels
...
Quantity
Pertinence
Manner
WeST – Web Science & Steffen Staab
Technologies staab@uni-koblenz.de
43. Hypertext Community
Linking people
Linking resources
Make linked data
understandable!
Linking data Linking stories
WeST – Web Science & Steffen Staab
Technologies staab@uni-koblenz.de
44. THANK YOU!
WeST – Web Science & Steffen Staab
Technologies staab@uni-koblenz.de
45. Literature
J. L. Austin. How to do things with words. Oxford University Press, 1962/1975.
M. H. de S. Bomfim, Daniel Schwabe. Design and Implementation of Linked Data Applications
Using SHDM and Synth. Int. Conf. Web Engineering 2011, pp. 121-136.
T. Franz, A. Schultz, S. Sizov, S. Staab: TripleRank: Ranking Semantic Web Data by Tensor
Decomposition. International Semantic Web Conference 2009: 213-228
Chierchia, Gennaro.; McConnell-Ginet, Sally: Meaning and Grammar : An Introduction to
Semantics. MIT Press, 1990.
Maximilian Nickel, Volker Tresp, Hans-Peter Kriegel: Factorizing YAGO: scalable machine
learning for linked data. WWW 2012: 271-280
E. Pietriga, C. Bizer, D. Karger, R. Lee: Fresnel: A Browser-Independent Presentation Vocabulary
for RDF. International Semantic Web Conference 2006, Springer 158-171.
S. Schenk, C. Saathoff, S. Staab, A. Scherp. SemaPlorer – Interactive Semantic Exploration of
Data and Media based on a Federated Cloud Infrastructure. In Journal of Web Semantics,
Elsevier, 7(4), 2009.
C. Saathoff, A. Scherp: Unlocking the semantics of multimedia presentations in the web with the
multimedia metadata ontology. WWW 2010: 831-840
R. Troncy, O. Celma, S. Little, R. García and C. Tsinaraki. MPEG-7 based Multimedia
Ontologies: Interoperability Support or Interoperability Issue? In Workshop on Multimedia
Annotation and Retrieval enabled by Shared Ontologies (MAReSO'07), Genova, Italy, December
5, 2007. http://www.eurecom.fr/~troncy/Publications/Troncy-mareso07.pdf
WeST – Web Science & Steffen Staab
Technologies staab@uni-koblenz.de