« Return to Thread: Agents in Prova rule language wrapped by UMO

Re: Agents in Prova rule language wrapped by UMO

by alex.kozlenkov :: Rate this Message:

Guys,

We have released the promised update to the Prova rule language, version 2.0 Beta 2F, that is now compatible with Mule ESB (see Prova News). The actual Mule-Prova gateway will implement a ProvaUMO interface and allow Prova agents to send and receive messages across the bus.

I am now going to clean up my gateway example and submit it in the contribution area. After that, I'm quite open for ideas and possible use cases. However, I understand that at this stage, I'm perhaps better positioned to drive this forward for awhile given my familiarity with Prova.

Given our mutual interest with JBoss Rules guys for interoperability, we are experimenting with various integration options that will integrate backward chaining, forward chaining, and Prova-style reaction rules under one umbrella. Rich Halsey is reporting on some progress integrating Prova with JEOPS and will be expanding that to JBoss Rules.

All in all, we are looking for ways of making various intelligent rule-based services available on the Mule bus.

Cheers,
Alex

Andrew Perepelytsya wrote:

Alex,

If you feel like releasing this module into a public domain, we can host it
in mule-contrib SVN repository, which is a repository for modules
'affiliated' with Mule, but not necessarily bundled with a core
distribution. Let me know if it sounds ok, we'll then arrange for access
permissions so you can commit to the repo directly.

Andrew

On 10/15/06, alex.kozlenkov <alex.kozlenkov@betfair.com> wrote:
>
>
> It took me a Sunday's work to wrap agents written in Prova rule language
> by a
> Mule UMO. Prova has rich built-in message-processing functionality based
> on
> reaction rules. Originally, we used local, JADE, and JMS protocols but it
> made perfect sense to use Mule now.
>
> Prova is now approaching release 2.0 and in the new beta 2f, this
> functionality will be enabled. This is an example of two agents:
> Agent001
>
> :- eval(agent001()).
>
> card('Mastercard').
> card('Visa').
>
> agent001() :-
> card(Card),
> sendMsg(XID, esb, "Agent002", request, card(Card)).
>
> Agent002
>
> rcvMsg(XID, Protocol, From, request, card(Card)) :-
> iam(Me),
> println(["Request received by ",Me," on protocol ",Protocol," from
> ",From,"
> about card ",Card]).
>
> The above starts Agent001 by running the goal "agent001()" then sends two
> messages for either of the two cards on the "esb" protocol. The second
> agent
> has a reaction rule that accepts messages matching the specified pattern
> and
> prints the main data.
>
> This is, of course, a toy example, but we could easily extrapolate this to
> useful rule-based decison making nodes. For example, the decision logic in
> the Mule Loanbroker example can be easily extermalised as rules. Prova is
> also good for various intelligent routing stuff based on Petri nets or
> state
> machines.
> --
> View this message in context:
> http://www.nabble.com/Agents-in-Prova-rule-language-wrapped-by-UMO-tf2448257.html#a6824157
> Sent from the Mule - User mailing list archive at Nabble.com.
>
>
> ---------------------------------------------------------------------
> To unsubscribe from this list please visit:
>
> http://xircles.codehaus.org/manage_email
>
>

« Return to Thread: Agents in Prova rule language wrapped by UMO