Panel: OWL Leaves the Nest Knowledge Integration for Ubiquitous Agents Harry Chen Image Matters LLC First International Symposium on Agents and the Semantic Web 2005 AAAI Fall Symposium Series
The Million Dollar Question What non-web application do you think will be the first to be deployed by industry and government? A knowledge integration application that uses SW languages to elevate the semantic meanings of the existing legacy data
Why Knowledge Integration Large amount of business knowledge (i.e., terabytes of data) will be stored in digital format, and not all of which will share a common data representation or data store. Majority of legacy data lacks explicit semantics, which makes difficult to build smart applications Valuable knowledge sometimes can only be acquired by reasoning over the hidden relationships between distributed data
Knowledge Integration in PerCom Building location-aware applications Instead of building a complete geographical model from the scratch, it will be more cost-effective to use SW languages to integrate legacy data from the existing data. Building smart sensor applications Instead of re-programming all sensors to emit data in a shared representation, it will be more cost-effective to use SW languages to map the existing representations into a shared semantic format.
Questions Are RDF and OWL the right languages for knowledge integration? How likely will developers adopt these SW technologies? It’s difficult for RDF and OWL to encode and use certain kinds of common sense knowledge that is essential for building smart applications. How can we address these issues?
Semantics is the Future Different technologies emerged in the recent months suggest the use of information semantics will play an important role in the future Folksonomy: Yahoo! MyWeb, Technoratic etc. Google Base: the truth is still out there… RSS/Atom/FOAF: MIT piggy-bank Commercial products: SemanticWork 2006, Oracle RDF database
SW Technology for Everyone People believe SW technology is too complex for software developers. I think not. The problem: we haven’t yet convinced people that SW technology can solve a real-world problem. “If we build it, they will come.” – not likely. “This thing will sell itself!” – don’t count on it. A solution: find a problem that SW technology can help solve, and do solve it well.
Problems in the Existing Technology Provenance is not supported by RDF and OWL Who said what It’s difficult to trace the pedigree of any inferred semantic web knowledge Why do you believe Harry Chen is located in Hyatt Regency Crystal City on 2005/11/04? It’s difficult to use RDF and OWL to encode and use certain common sense knowledge. “Find me a nearby gas station” “Is it faster for me to take subways or drive to work?”
KnowledgeSmarts TM KnowledgeSmarts TM is a middleware system that uses Semantic Web technology to enable knowledge integration Help agents to dynamically query heterogeneous data from multiple legacy data stores via unified RDF models Help agents to track the provenance of asserted and inferred RDF triples Provide extensible ontologies and built-in logical inference engines for spatial and temporal reasoning Provide an event-driven process execution engine for monitoring changes in the underlying data stores
Summary Knowledge integration applications will be the first SW applications to be deployed by government and industry. Ubiquitous agents will greatly benefit from the realization of semantic knowledge integration – e.g., reuse and share existing legacy data People will adopt SW technology if we can show such technology can solve real-world problems KnowledgeSmarts TM => knowledge integration
Backup Will the use of SW languages drive a unified web-based design in the future mobile computing system? Will the impact of RDF and OWL on the systems and communication ultimately be great than the World Wide Web?