The Pragmatic Web: Preliminary Thoughts Munindar P. Singh

advertisement
The Pragmatic Web:
Preliminary Thoughts
Munindar P. Singh
www.csc.ncsu.edu/faculty/mpsingh
singh@ncsu.edu
April 2002
Conclusions
When the Semantic Web comes around, it will
come in the form of the Pragmatic Web.
• Need to develop new abstractions, not
simply rehash old work.
• Must consider (besides data)
– Process.
– Interaction.
– Context.
Semiotics
• Syntax: structure.
• Semantics: meaning as captured in syntax.
– Function from structures to “meaning objects.”
– Compositional semantics: above function is a
homomorphism.
• Pragmatics: meaning in usage.
– Context.
– Indexicals, e.g., I, here, now.
– Deixis, e.g., this, that.
Semantic Web
• Mission: enable information to be shared by
applications across the web.
• Methodology: Develop standardized
representations for markup.
Does the methodology serve the
mission?
• Why will standardizing the ontology
representations solve the problem?
– Though valuable, ontologies didn’t solve the
problem of sharing information sharing in
databases.
– Standardizing representations for them will
simplify exchange of metadata, but not of its
usage.
Challenges for the Semantic Web
• Technical:
– Limits of semantics: designing generic representations
independent of usage.
– Engineering: Easy to build demos; difficult to build real
systems.
• Technical culture: emphasis on visions that are
– Glitzy: mobile or pervasive devices talking to each
other to enact fancy demos, conceptually glued together
for a demo
– Weak on identification of anything new, except that
they execute over the Web or with new devices.
Service Composition
• Essential to personalize services.
– The value is in the context.
– No need for intermediaries otherwise, e.g.,
airlines versus travel agents.
Abstractions for Openness
All science is either physics or stamp
collecting – Ernest Rutherford.
• Old physics:
– Describe process of composition procedurally
in a trivial execution model (a graph).
– All subtlety pushed to the data connectors.
• So what is the new physics?
Describing Services and Desired
Compositions
• Properties of services (inadequacy of
WSDL):
–
–
–
–
Autonomy.
Heterogeneity.
Long-lived, evolving.
Cooperativeness.
• Specifying compositions.
• Testing compliance.
Discovering Services
• Inadequacy of UDDI.
• Trust.
– Dependent on usage context.
– Often without trusted third parties.
• Evaluation.
• Reputation.
Describing Compositions
• Inadequacy of WSDL in supporting
compositions.
• Inadequacy of WSFL, DAML-S for
specifying compositions (they have other
features and they are improving, but still).
– Rehash of workflow: procedural primitives
(split, join, etc.)
Principles
• User before provider. Consumers are presented as
composers of services with idiosyncratic needs
and context dependent requirements.
• Interaction before representation. Services should
describe the interactions and minimize the
representational dependencies.
• Process before data. Processes behind the data
give a better clue to its meaning than a processindependent (impossible) semantics.
Central position to process and context, not to data.
Are we arguing against the
Semantic Web? No
If man evolved from monkeys and apes, why
do we still have monkeys and apes? –
George Carlin
• Work on pragmatics will promote the
semantics, just as the semantics promotes
the syntax.
• The pragmatics endeavor is a special
direction within the semantics endeavor.
Plugs
• IEEE Internet Computing
– Theme issues.
– Always welcomes solid articles on Semantic
Web and Web Services.
• Workshop on trust, deception, fraud,
privacy, 2002.
• Workshop on regulated agent-based social
systems, 2002.
Download