by James Hendler

Something hit me the other day — in late 1998 I went to DARPA and took over the “Control of Agent Based Systems” (CoABS) program.  This program focused on developing middle ware for agent-based systems, essentially an interoperability program aimed at making different agent architectures work together.   At the time, there were a number of different agent frameworks being explored, and several relatively successful architectures - the biggest problem was they didn’t play well together, the second was that there was a lot of work involved in getting them to communicate because every application had to be engineered from scratch, and none of the knowledge could be reused.  The first program was addressed by the “CoABS grid” - which I’ll return to in a minute - the second was the genesis of the DAML program — the darpa AGENT markup language.  DAML led to the successful development of a shared KR language, DAML+OIL, which formed the basis for the Web Ontology Language OWL which became a standard in 2004, and which has had a great deal of development and even some commercialization to date.

The CoABS grid focused on being a middleware level that would make it easy for agent applications to communicate together and for legacy systems to be wrapped to communicate like agents.  The program was based on JINI, which at the time was clearly the best choice.  Nowadays, while the Grid is still in use in some military programs, the middleware concept behind the software has become much more popular in the Web services framework.  Instead of needing proprietary technology to wrap your legacy, there’s SOAP, WSDL and a stack of service languages, and these also include what is needed to register services and to do many of the other things the grid did.
So we have services that now provide a standard for interoperability between systems, easily exploitable by agent-based systems.   We have RDF and OWL which provide a KR language that is certainly capable enough for many agent applications.  We have tens of thousands of ontologies that make domain engineering easier.  We have many large Web providers that make access to their systems available through some sort of service interface or in easily wrappable ways.  We have XML described databases, as well as SPARQL endpoints making data query for agents more accessible.

What we don’t seem to have is agent-based systems in any serious way!  AAMAS proceedings seem to have very similar papers to what was being published a few years back, the applications track has a fair amount of interesting work, but very few papers that actually talk about any sort of deployment, and certainly nothing in the wide-spread use that was predicted back in the late 90s.  AgentWeb seems to have last been updated somewhere around 2005, and the bulk of the papers I can find published in the last year or so are filled with all kinds of wonderful theory, just not any agents.  It’s hard to even find new papers on agent architectures.  Prior to 2000, IEEE Intelligent Systems ran two special issues on agents and had lots of papers in between, now we hardly get any and I haven’t seen a special issue proposal in this space since I took over as EIC
So what happened?  Semantic Web work has continued great guns, and with systems like Joost and RadarNetworks using the technology, it’s hard to argue that the technology is not being deployed in Web scale applications.  But agents, were one of the motivations for all this,  and they were at least the one Ora Lassila and I had when we started working on the Semantic Web article in Scientific American (May 2001- it’s been a while!)).

So I ask the question in the title — where are the agents?

— Jim Hendler,

This entry was posted on Thursday, March 8th, 2007 at 9:57 pm and is filed under Software, Personal Ramblings, WSRI. You can follow any responses to this entry through the RSS 2.0 feed. You can leave a response, or trackback from your own site.