- From: Chiusano Joseph <chiusano_joseph@bah.com>
- Date: Sun, 19 Sep 2004 13:43:14 -0400
- To: Drew McDermott <drew.mcdermott@yale.edu>
- CC: public-sws-ig@w3.org
Regarding automatic composition: I've been keeping up with a lot of the wonderful work that is going on not only in this online community but also some of the additional research that is being done for the automatic composition and invocation of Semantic Web Services (Sheshagiri and others). I'm beginning to see an area that I view as a potential gap, regardless of how advanced the work on automatic composition and invocation of Semantic Web Services becomes, and I wonder if I might perhaps be overlooking some references. The area in question has to do with semantically aware relational databases, and the ability for a Semantic Web Service to request and obtain automatically the information it requires from a relational database - which is separate from, but related to, automatic composition and invocation of Semantic Web Services. Here is a scenario: <Scenario> Suppose we have a Travel Agent Web Service (classic example), and that Travel Agent Web Service, among other things, dynamically discovers Hotel Reservation Web Services, trying to find the best bargain(s) for the traveler. Suppose also that the Travel Agent Web Service accepts input from a "traveler profile" that is a record in the travel agent's relational database. This profile lists, in addition to basic information such as name, address, etc., various preferences of the traveler (i.e. the features that they desire for a hotel). The Travel Agent Web Service would, of course, use this information when discovering Hotel Reservation Web Services. Suppose now that there is a Hotel Reservation Web Service that meets various criteria (to include the traveler's preferences), but it needs to know additional criteria - i.e. criteria that are not carried "on the wire" for the request (e.g. number of children). Let's also suppose that this information is included in the traveler profile, but (for whatever reason) was not passed on the wire. The Travel Agent Web Service could simply not discover this Web Service because of the criteria mismatch - or, it could discover it and handle the mismatch. Handling the mismatch could entail an action by either the Hotel Reservation Web Service or the Travel Agent Web Service to request that additional information from the Travel Agent relational database (let's assume that security settings will not allow the Hotel Reservation to make this request successfully because it is not a trusted entity, so the Travel Agent Web Service will do so). At this point, we need the following to happen: (1) The Hotel Reservation Web Service must relay to the Travel Agent Web Service the information that is missing, and (2) The Travel Agent Web Service needs to obtain that missing information from the Travel Agent relational database It is #2 above that I perceive as a current gap - i.e. unless the Travel Agent relational database is sufficiently "semantically aware" (i.e. perhaps it implements an OWL ontology whose classes and properties are mapped to the database tables/fields respectively), there is no efficient and accurate way that the required information can be obtained from the Travel Agent relational database. </Scenario> So my assumption is that no matter how well the automatic discovery, composition, and invocation of Web Services works on a real-time basis, unless we have semantically aware relational databases that can handle requests automatically as described above, we will never realize the full potential of Semantica Web Services. Is this assumption valid? Is work being done in the area of semantically aware relational databases (particularly from the largest database vendors) that I may not be aware of? Kind Regards, Joe Chiusano Booz Allen Hamilton Strategy and Technology Consultants to the World Drew McDermott wrote: > > > [Sam Watkins] > > > > Obviously, this requires the author of the mediator to be fluent in = > > terms of both ontologies in order to conserve meaning. > > I think it is interesting to find the means of constructing these = > > mediators without the author having to be fluent in OWL-S (or any other = > > ontology description language) but just in the meaning of the = > > ontologies. > > It's difficult to see what you envisage. > > > While this would allow assisted composition using mediators, rather than = > > fully automated composition, I think this would be a good stepping stone = > > towards automated composition. > > Fully automated composition would presumably require the computer to > be as smart and knowledgeable as we are. That's not going to happen > in the foreseeable future. > > -- > -- Drew McDermott > Yale Computer Science Department -- Kind Regards, Joseph Chiusano Associate Booz Allen Hamilton
Received on Sunday, 19 September 2004 17:44:05 UTC