Index of /CVS_CONTENT/moby-live/Docs/OntologyDevelopment/

README # # Revision History # # 15/4/2004 Simon Twigger (simont@mcw.edu) # INTRODUCTION MOBY is a system for the dynamic discovery of bioinformatics services available on the WWW. In order to effectively find a service that performs a particular task, an accurate, machine readable description of that service is required and this is best achieved through shared vocabularies or Ontologies. The current MOBY service ontology is very sparse, consisting of 6 major classes of service which was deemed woefully inadequate to efficiently describe the wide array of bioinformatics and related services available. It was decided at the April 2004 MOBY developers meeting at Cold Spring Harbor that some effort should be put into expanding this ontology and this is the start of that work. Existing service ontologies such as those developed by the MyGrid group demonstrate how others have addressed this issue (See the MyGridDocs/ subdirectory for DAML/OIL versions of this ontology, try opening with OilEd editor http://oiled.man.ac.uk/). Similarly, large collections of bioinformatics services such as those offered by Emboss and Pise give a view of the breadth of services that exist now and hence what a service ontology should minimally encompass (http://bioweb.pasteur.fr/intro-uk.html). Complicating factors as the MOBY ontology develop are: * the desire to avoid reinventing the wheel and contributing to the proliferation of redundant ontologies * the choice of format for the ontology itself - XML/RDF, DAML/OIL, OWL, OBO, (your favorite format here) * the impact of using LSIDs (Life Science Identifiers) as a way to find resources on the web and potentially tie together separate ontologies * the desire to have something workable in a reasonable time frame. As a start, I've begun assembling the ontology in OWL using the Protege2000 editor - (http://protege.stanford.edu). It is fundamentally a simple hierarchy of services and doesnt currently use the power of OWL to tie together different classes. However, this would be an attractive feature as many services cannot be classified into a single class and this could be addressed using the more complex structure of OWL in combination with a Reasoning engine such as Racer. This would require a fundamental change to MOBY-S though it would be very much in line with the move to S-MOBY... --