Semanticweb.com recently caught up with Dr. John Domingue, deputy director of the U.K.’s Open University’s Knowledge Media Institute. Dr. Domingue specializes in researching how semantic web technology can support the automated creation of web applications from web services. KMI was set up at the Open University -- which was created in the 1960s to serve those who missed out on higher education and now has 220,000 students studying via distance learning courses -- to carry out research related to the creation and sharing of knowledge, and today one of its big research topics is the semantic web.
Currently, Domingue is running about five EU projects in the KMI Lab on the semantic web, with a combined value approaching $50 million.
Semanticweb.com: Tell us about your specific work in the semantic web field, and what you research is based on.
Domingue: My specific area is in semantic web services, applying semantic web technology to web services, with the angle of automatically constructing applications.
All of the research is based on our conceptual framework and implementation platform. The framework is WSMO -- the Web Services Modeling Ontology. The main things about WSMO is that it leverages four top principles -- everything is underpinned by ontologies; there are web service descriptions where we differentiate between the function of a web service and how you invoke it; and there is the concept of a goal. So let’s say that me, as a user, my goal is to go on holiday in southern Europe in a child-friendly place -- you specify that in the goal. Then web services will match that goal -- like, I am a hotel available for certain dates, and I am an airline company, so we match between the user goal and services -- you combine the services to reach the goal.
So the elements are web service composition, invocation and discovery, and also a top principle is the concept of a mediator. You specify your goal and I grab some web services in the wild on the Internet, and there will be mismatches in the structure of the data, the underlying conceptual model, in requirements (like you must send a credit card for a service). We have a language for describing mediators that fix these types of problems.
Then the implementation platform is called IRS-III. That is a platform we constructed where we can manipulate and run WSMO-based models. So you specify the goal, send it to IRS, and it acts as a broker between your goal and web services.
WSMO started in an EU project called DIP. that ran from January 2004 to the end of December 2006.
Semanticweb.com: Why do we need to apply semantic web technology to web services?
Domingue: Scalability, that’s the line we always put in. Web services as they are currently won’t scale, because every time you want to make a change a human software developer is involved in the loop. As you move to billions of services you need to automate some of these aspects and the only path to automation is to describe some of the components using semantics.
Addtionally, because web services correspond to business services we can turn organizations from monolithic black boxes to a set of micro functionalities which add value for the customer. If I can describe the services offered semantically, and they can be recomposed on the fly automatically and that’s very exciting.
We don’t require that web services change at all. This is a layer on top of web services. For web services you have an end point, a URI, to send a message to. Then some other person describes that semantically. One possible complaint is that the description may be complex. A commercial company may not want to learn about this ontology and heavy-weight stuff. What I say to them is look at the new W3C recommendation on SA WSDL, which is semantic annotations for WSDL (Web Service Definition Language). If a company wants to make a first step to semantics, then they can look at that -- put a hook into a WSDL file that points to some semantic definition.
Featured in the
IE 7
Firefox 2.0
