I’ve just read a Tech Report which advocates the use of the Web Ontology Language (OWL) as a description language for service discovery in mobile computing environments. How the authors think this could possibly work is beyond me. These guys have written papers on this which have actually been accepted to conferences. Their description language is based upon previous work they have done: DReggie. I understand that an ontology based solution might work for DReggie, since it is an extension to Jini, and therefore relies on a central lookup service. Service descriptions in queries and advertisements could be validated against schemas maintained at the lookup service. But how is this transferrable to the domain of truly mobile computing, where it is infeasible to expect every node to maintain schemas defining the myriad of service types it will come across in the course of processing advertisements and queries? This is perhaps the least of my worries. OWL comes in a number of flavours (OWL Lite, OWL DL and OWL Full). But I contend that even OWL Lite is too heavy to be used in a mobile computing environment. For a start it’s based on RDF. During honours I had lots of experience with RDF while analysing CC/PP. It’s bloated, big time. Further, instances need to be validated. Then, to compensate for the general bloatedness of the OWL family of languages, OWL Lite imposes some silly limitations like only supporting cardinality values of 0 or 1. I mean, what the hell kind of stupid limitation is that!? To round things off nicely, the acronym doesn’t even fit (WOL?). Given that it would be next to impossible to do proper validation in an ad hoc mobile network, what is the point of using OWL (or anything based on RDF for that matter?). I’d like to see them try to run an RDF validator, let alone an OWL inference engine such as their own F-OWL.
On the contrary, my non-OWL based description language does run on a phone. No, descriptions are not validated against schemas. There is no inference engine. But then, I contend that these processes are not really feasible for pervasive computing environments consisting of light-weight mobile nodes. The whole world has gone OWL crazy, and not stopped to think about it. From now on, I dub my own description language AWOL: the Anti-Web Ontology Language. And the acronym actually works.
