opportunistic reasoning for the semantic web: adapting reasoning to the environment carlos pedrinaci...

Post on 27-Mar-2015

217 Views

Category:

Documents

1 Downloads

Preview:

Click to see full reader

TRANSCRIPT

Opportunistic Reasoning for the Semantic Web:

Adapting Reasoning to the Environment

Carlos PedrinaciTim Smithers and Amaia Bernaras

The Semantic Web

• After 10 years research on the Semantic Web has already produced a considerable amount of technologies

RDF

OWL

WSMO

OWL-S

Sesame

Jena

Pellet

KAONKowari YARS

IRSWSMX

AqualogFlink

eMerges

DBin

So Far, So Good But…

• The main focus is essentially on the development of tools that generate data represented in Semantic Web languages

• Applications are mainly limited to query engines and information aggregation

So Far, So Good But…

• Where is the killer application?

• Which are the distinguishing features with respect to Web 2.0?

• Why are companies somehow reluctant to embracing the Semantic Web?

• Indeed, the difficulties are not just technical but we need to get these right!

Some Technical Issues

• Scalability– The Web is huge (and is growing)

– Reasoning is expensive

– Need for further expressivity which makes things worse

– Knowledge Engineering is particularly challenging

Some Technical Issues

• Completeness– Not possible within such a dynamic

environment

– We need to reason in an opportunistic and incremental manner

• Consistency & Correctness– Not realistic for the Web

– We need non-monotonic reasoning and truth-maintenance mechanisms

Some Technical Issues

• The Web is essentially dynamic

– The previous problems just get worse

• Worst of all, a Semantic Web application should deal with all these issues, plus the typical Engineering challenges!

Let’s Shift the Focus!

• Let’s view the Semantic Web as a phenomenon emerging from the interaction of intelligent applications over the Web and not as an entity in itself

• In this scenario Scalability, Expressivity, Completeness, Consistency, Trustworthyness more manageable issues

Opportunistic Reasoning

“Ability of a system to exploit its best data and most promising methods”

Erman et al. 1988

Blackboard Model

Blackboard Characteristics

• “Divide and conquer” - Problem-solving expertise partitioning– Simplifies Knowledge Modelling– Promotes reuse– Applicability of diverse representation

and reasoning techniques

• Collaborative and concurrent reasoning

• Event-based reasoning

Blackboard Applicability Criteria

• A large solution space• Noisy or unreliable problem data• A continuous data flow• The need to integrate diverse and

heterogeneous data• The need to integrate different

sources of knowledge

Blackboard Applicability Criteria

• The need to apply several reasoning methods

• The need to develop various lines of reasoning

• The need for incremental reasoning

• The need for an opportunistic control of the reasoning process

Blackboard Applicability Criteria

• The need for an event-based activation of the reasoning

• High complexity of the task• The need for a mixed initiative

where computer and users can interchangeably take the initiative

• Meta-reasoning or conscious reasoning– Drive & Explain the reasoning process

General Applicability

General Applicability

General Applicability

General Applicability

Applicability to the Web

• Particularly well-suited for the Web– Opportunistic Reasoning

• Adapted to the dynamism of the Web

– Flexible and versatile• Wide applicability, seamless integration of diverse

reasoning engines, languages, and tools

– Modular• Maintainable, extensible

– Distributable• Computation distribution• Adapted to the Web (Ontologies, URIs, Web

Services)

An Infrastructure…

Several Applications…

Web-based Events Design Support

System

Music Rights Clearing

Organization

Conclusion

• Considering the Semantic Web as an emerging phenomenon has important pragmatic consequences

• Opportunistic Reasoning seems particularly appropriate for reasoning over the Web

• Appealing characteristics from an Engineering perspective that make it a good candidate for supporting Semantic Web applications

Thanks

Thank you for your attention

top related