Presentation on STARLab's research, the GOSPL method and prototype presented at the second IFIP WG12.7 "Social Networking Semantics and Collective Intelligence" workshop in Amsterdam (26-27 April 2012).

1. +
Grounding Ontologies
with Social Processes and
Natural Language
2012-04-26
IFIP WG 12.7 Workshop #2

2. +
Definition of Ontology in
Computer Science
n A conceptualization is a mathematical construct that contains
abstract references to (1) objects, (2) relations, (3) functions,
and (4) events as may be observed in a given real world.
n An ontology is a shared, [first order] logical, computer-
stored, specification of such an agreed explicit
conceptualization.
n [Tarski 1908, Gruber 1993, Studer 2000, et al.].

3. +
Definition of Ontologies in
Computer Science
n In summary: Semantics = Agreed Meaning
n Links symbols in autonomously developed systems to shared
reality
n Agreed among humans as cognitive agents
n Stored in ontologies
n key technology for interoperability
n ontologies ≠ data models, but provide annotation for them
n support both human- and system-based reasoning

4. +
Tri-sortal Network of 3 Networks of
Actors

5. +
Interoperation != Integration
n The autonomous nature of actors needs to be respected
n Interoperation stems from a need or wish to communicate,
and collaborate
n à Motivates the need for agreements, contracts and the
meaningful exchange of concepts

6. +
The need for dual perspectives
n Human perspective: high level reasoning about “shared”
concepts
n put humans “in the loop”
n natural language contexts
n System perspective : vocabulary agreements, lexons
n large volume data access
n low level reasoning

7. +
Ontology Engineering Methods:
Learning from Databases
n Technology matures: involve the less IT-gifted IT experts
n Natural language discourse analysis (NIAM, ORM) as used for
databases
n Use legacy data / output reports / interviews, abstraction
into fact types
n Lift data models into ontologies, remove application-specific
context

8. +
Developing Ontology-Grounded
Methods and Applications
n Communities of users / domain experts own the ontology.
Make use of discourse, social process and “legacy” resources
n Ontologies as approximations of perceived reality at type
level! As ontologies evolve, they approximate the real world
n Users / domain experts rule at every step
n Facts holding in a certain context (the community, see later)

9. +
DOGMA
“Double Articulation”: Ontological Commitments in DOGMA
Lexon Base
Commitment Layer
Applications

10. +
Commitments in DOGMA
n Commitment = < Selection, Encoding, Constraints >
n Where Selection = set of lexons with various Context-ids
n Encoding = reference mapping: Application symbols to lexon
terms
n Constraints = set of Ω-RIDL* statements (expressed in lexon
terms)

11. +
Towards Hybrid Ontology
Engineering
n Revisit discourse analysis, pragmatics, semiotics
n Model communities as 1st class citizens
n Formalize methodologies based on NL involvement of
domain experts à Revisit discourse analysis, pragmatics,
semiotics
n Upgrading role of legacy systems in enterprises
n Scalable semantic re-exploitation of RDF and LOD resources

12. +
Grounding Ontologies with Social
Processes and Natural Language
n Hybrid Ontology Description (HOD) HΩ=<Ω,G>
n Ω is a DOGMA Ontology Description (Lexon base, commitments
and a mapping from terms to concepts)
n The contexts in hybrid ontology descriptions communities
n G is a glossary, a triple with components
n Gloss, a set of linguistic, human-interpretable glosses. Mappings
from community-term
pairs or lexons to glosses

13. +
Method
Implementation of the ontology
OWL, RDF(S), …
E.g., with tools offered by the RDB2RDF community such as D2R Server.
Semantic Interoperation of IS through
Formalized Social Processes
03/21/12 15

14. +
Lexons + Constraints

15. +
Method
Manage Articulate Create Constrain
Community Commit
with glosses Lexons Lexons
Manage Semantic
Interoperability Gloss-
Synonym
Requirements Equivalence

16. +
Discussion oriented + Traceability

17. +
Exploiting the annotated data
(in RDF)

18. +
Gloss Driven!

19. +
Joint work with CVC on Ω and MTB
Co-evolution

20. +
Exploiting RDF thanks to Hybrid
Ontology Implementations
n Augmenting RDB2RDF
Mappings by means of Ω-RIDL
Commitments
n Adding semantics to the
database structure

21. +
Exploiting RDF thanks to Hybrid
Ontology Implementations
n Fact-oriented querying of RDF.
n LIST Artist NOT with Gender with
Code = ‘M’
n In SPARQL:
SELECT DISTINCT ?a WHERE {
?a a myOnto0:Artist. OPTIONAL {
?g myOnto0:Gender_of_Artist ?a.
?g myOnto0:Gender_with_Code ?c. }
FILTER(?c != "M" || !bound(?c)) }