Comments on: Interesting trend: CEP at BRForum, ORF, RuleML events

By: Adrian Paschke

Adrian Paschke — Sat, 27 Jun 2009 10:07:27 +0000

I would like to point to RuleML-2009 (http://2009.ruleml.org) as one of the major conferences addressing rule-based CEP / reaction rules.

There will be a special track on rule-based Event Processing and Reaction Rules on RuleML-2009 (see the call for papers) and a keynote about CEP this year.

CEP has been a topic of the RuleML conferences since the very begining, even before the term CEP was officially “born”.

Here some historicaly references (http://2009.ruleml.org/past) which might be of interest:

- Keynote by Opher Etzion “Towards an Event-Driven Architecture: An infrastructure for event processing” on RuleML-2005

- Special workshop on Reaction Rules and Event Processing at RuleML-2006

- Session on Reaction Rules and Rule Applications at RuleML-2007

- Keynote by David Luckham on “The Power of Events: An Introduction to Complex Event Processing in Distributed Enterprise Systems” at RuleML-2008

- Keynote by Paul Haley on “Event and Process Semantics will Rule” at RuleML-2008

… and many interesting papers about reaction rules / rule-based CEP in the conference proceedings of the past years (Springer LNCS and IEEE proceedings), as well as demos in the RuleML Challenge.

Cheers Adrian

By: Paul Vincent

Paul Vincent — Tue, 16 Jun 2009 15:36:17 +0000

“Semantics in CEP” is the theme at The 8th International Conference on Ontologies, DataBases, and Applications of Semantics (ODBASE 2009)- http://www.onthemove-conferences.org/index.php/odbasehornav - in Portugal. This academic conference conflicts (datewise) with BRForum and RuleML09.

By: Paul Vincent

Paul Vincent — Thu, 04 Jun 2009 10:39:40 +0000

Hi Peter: good luck with your project. FYI the BE Query Language also uses Rete…
Cheers

By: Peter Lin

Peter Lin — Thu, 04 Jun 2009 04:26:52 +0000

I agree there are multiple classifications. I’ve been toying with the idea of a Stream RETE engine, which i plan to implement this summer for fun. I plan to write up my findings and share the results with the broader business rule community in the hopes of pushing the state of art.

peter

By: vincent

vincent — Tue, 02 Jun 2009 10:18:31 +0000

Hi Peter: I guess there are multiple classifications and characteristics of applications, leading to different patterns:
- event driven vs event processing
- #events vs #rate of events to be processed
- #events vs #temporal rules and relationships to be processed

I’m guessing that *some* temporal Rete approaches would be ideal for large# temporal rules and relationships, but less so for #rate of events. Indeed for BE we have some simple architectural pattern changes to adopt to some of these different application characteristics.

Cheers

By: Peter Lin

Peter Lin — Sun, 31 May 2009 22:07:29 +0000

I was thinking on a general temporal logic perspective and not specific to what drools does. It sounds like Tibco BE doesn’t go the temporal operator route either. The slow performance of Allen’s approach isn’t fatal. Though ultimately, I wouldn’t want to use his approach in a performance sensitive application.

By: Paul Vincent

Paul Vincent — Fri, 29 May 2009 19:21:31 +0000

Hi Peter: are you referring to Edson’s ORF paper on DROOLS? If so, the proof of the pudding will be in the eating - and IMHO the performance will probably be “good enough” for many prototypes and applications. Remember there are plenty of ways to skin the temporal cat…

Talking of temporal cats: I’ll have to explain TIBCO’s approach in my ORF paper or future blog post. I wouldn’t describe it as “temporal rete” - remember TIBCO BE uses multiple models to represent problems, not just rules (i.e. state models / continuous queries / timer rules etc).

At some point EPTS is probably going to cover temporal operations in its Interoperabilty working group - it would be good to have the DROOLS gang and people like yourself with an interest in ‘CEP using rules’ consider contributing…

Cheers

By: peter lin

peter lin — Fri, 29 May 2009 13:33:13 +0000

I’ve been looking at allen’s approach to temporal logic for the last 2 years and have come to the conclusion his approach isn’t practical. There are significant flaws in his model, which 3 other papers point out. Aside from that, the performance of his approach is too slow. The third issue is the execution model proposed by Allen is not compatible with RETE network.

I’m curious, does Tibco plan to publish a paper on how BE handles temporal logic? Even though my rule engine jamocha is just a research project, it supports a limited form of temporal logic through temporal facts and temporal patterns. This summer I plan on implementing stream Rete engine, which will hopefully help everyone progress towards a general purpose model of temporal logic.

peter