How to Develop the Best Semantic Services?
The "We Know Best" Approach
Really smart people develop formalisms in a vacuum
We did this with software agents: KQML and FIPA
And with OWL-S, WSMO (UDDI is a different kind of failure)
Same people invent applications to match formalisms
Results are papers presented to each other
The "Who Does Best?" Approach
Really smart people develop test services and applications
Other really smart people tryout their formalisms
Results are implementations analysed by each other
Formalisms, services, and applications are open
The evaluation is based upon software engineering: how hard was it for systems to adapt to new services and problems?
© 2007 Charles Petrie
- permission to reproduce widely with attribution
