Let me articulate a slightly different point of vew. In much of the
discussion on Jess Modules, people have talked about the case where a rete
engine has a bunch of knowledge modules, but there's just one module that
needs to be the "focus" module. This, of course, caters to the scenario
where the "system very naturally partitions itself into a number a
sequential phases".
But how about the case where you want to concurrently run, say, 200 agents -
with each agent having it's own, independent knowledge base. Conceptually,
each one of these guys needs to have its own rete engine. In an actual
implementation, though, you might want to run all 200 on the same rete
engine. From an OS point of view, these agents (providing, say, a service to
users behind a web server) are going to be asleep a whole lot of the time.
So you definitely don't want to spawn 200 rete engines.
Modules might help here by partitioning the working memory and the rule base
into multiple namespaces, with facts being asserted and rules firing
concurrently in each namespace, without any one namespace having the
"focus". Each agent would be free to assert facts (and even to to change
rules) within it's namespace without fear of stepping on another agent's
toes.
In other words:
Somebody on the list said - "However, the primary difficulty, I think, is
not in maintaing the autonomy of the agents, but in providing the
'lightweight' communication of facts between agents. Adding modules to Jess
would be one approach".
How about using modules to provide autonomy for agents within a rete
engine - thus providing greater scalability for the system by allowing you
to have many more agents (servicing many more users) running.
- Lalit
---------------------------------------------------------------------
To unsubscribe, send the words 'unsubscribe jess-users [EMAIL PROTECTED]'
in the BODY of a message to [EMAIL PROTECTED], NOT to the
list. List problems? Notify [EMAIL PROTECTED]
---------------------------------------------------------------------
