The Grid needs ontologies - onto-what?Brussels 12 December 2002 The Grid is not only about accessing remote computers, it is as much about accessing remote data in large quantities. But it would be very time consuming to figure out for each database you may need, what is in it, what is the value of the information, where it fits into the whole knowledge world and how you can access it. This is where web and grid services come in, this is where the semantic web comes in, this is where ontologies are needed: standard ways to capture knowledge. In the end this will lead to the Semantic Grid. This all was discussed at the "Convergence of Web Services, Grid Services and the Semantic Web for delivering e-Services" conference in Brussels, December 12, 2002. That is a lot of buzzwords. But underneath are a few important concepts that are useful for Grid people.
Grid computing has two aspects that have it differ from older meta-computing and distributing computing efforts. First the scale of the data handled. Grid computing is not only about accessing computing resources, but also large data sources, like virtual laboratories with experimental data capturing equipment as microscopes or telescopes, but also stored data sources like medical information and biological information.
The other aspect is the use of computing sources and data sources that are not controlled by the user or his organisation. In eScience, for instance, one wants to combine a medical data source in the US with a special microscope in Japan steered from Europe, as was shown at iGrid2002 in Amsterdam. Or a company needs massive peek computing power to perform an analysis. In these cases one needs mechanisms to get services delivered in time and within budget.
To achieve these two aspects, one needs:
At the conference in Brussels, Carl Kesselman, Director of the Center for Grid Technologies, Information Sciences Institute, University of Southern California, gave his view on the evolution of the Grid. Central to his Grid concept are communities of people committed to a common goal. In science, brain specialists are an example as are high-energy physics. They consist of teams with heterogeneous members with different capabilities. There is not one organisation that has all the required resources or skills and team members to be distributed around the globe. Hence, the Grid should enable resources sharing and usage co-ordination in dynamic multi-institutional virtual organisations. In the non-Grid world, web services are emerging as a standard to access and account loosely coupled computing services. Although there are a lot of open ends to web services and an avalanche of competing standards, it does not seem worth for the Grid community to propose just another web services standard. Hence the Open Grid Services Architecture (OGSA) approach of adding on top of web services. OGSA advocates a service orientation to virtualise resources. From web services standard interface definition mechanisms and protocol bindings are taken. From Grids, the service semantics and life cycle management. But, given a set of services, how do we do a better job of finding out what services we want to use? How can we better configure services? How can we better compose and nest them? According to Dr. Kesselman, we need to find ways to better present services. We can capture information via structure, X.509 certificates, schema and meta-data, and we can express knowledge as relationships between entities. Add that up and you have an ontology vocabulary of the Grid: a general formalised set of information and statements about things on the Grid. As regular Primeur/EnterTheGrid readers know, we at Primeur/EnterTheGrid are already implementing a small part of this big Grid ontology vocabulary with http://EnterTheGrid.com - the largest Grid related catalogue in the world, for nearly two years now, using basic standards like XML, MPEG-7 and vocabularies for, for instance, descriptions of TOP500 and other supercomputers and classification schemes for, for instance, Grid architecture and Grid usage. Dr. Kesselman takes it all one step further, asking to where this all leads. He predicts the Cognitive Grid, called Semantic Grid by others, where Grid Services, Ontologies and Knowledge Driven Services will all have their place. But that will take many years to happen. For now, just remember the buzzword "ontology": a way to capture and present in the computer, knowledge all people in a certain community have.
Ad Emmen |
 |