Prof. Adrian Paschke has posted the presentation made by himself, Prof. Harold Boley and myself on Semantic Complex Event Processing at SemTech this year (previously blogged about here and here) - and makes an interesting comparison with a prior public presentation in this area 2 years ago. Semantics in CEP were also one of the research topics at DEBS this year, too…

UPDATE: I’ve had a few requests for non-slideshare versions - so here is the PDF .

Elisa Kendall from Sandpiper and Dr Mohammad Ketabchi from Progress were my honorable partners on the Semantic CEP Panel at SemTech today. Q&A covered some interesting questions:

- isn’t the event pattern - decision - reaction view too simplistic?

This was complaint about the “simplified view of the world” slide I presented. Of course, being “simplified” it did not cover, for example, possible multiple levels of abstraction or feedback loops, while learning and update mechanisms are certainly still applicable.

- managing explosion of outgoing events - won’t reaction events potentially overwhelm an event bus?

Usually CEP is “reducing” numbers of simple events (observations etc) into a fewer number of complex / business events - so while this could be a problem, it tends not to be in most practical applications and/or can be managed through the middleware layer.

- what next for financial event processing?

Increased regulatory compliance rules will be applied to more financial operations and transactions by both banking and government agencies… probably a growth area for CEP technologies!

- surely semantic is not just limited to “static” ontology definitions - consider for example ontologies of actions and events?

This was against in introductory slide showing semantic community focus with the “FOAF” type logical relationships - good for text search problems and such, but less so for business operations and behavior - processes and services… Elisa also commented that dynamic classification was certainly a semantics capability, but was more a research topic in “government applications” rather than a productised capability right now.

I was reminded of the importance of semantics at the coffee break at this morning’s Semantic Technology Conference when I watched a fellow attendee nearly succeed in unscrewing the tap on the coffee urn (instead of simply pulling the tap). Meaning is everything. Meanwhile, Harold Boley, Adrian Paschke and myself are presenting this afternoon on the possibilities of Semantic CEP. This morning 2 of the W3C RIF co-chairs presented on the W3C Rule Interchange Format and other rule standards, to a good crowd that seemed at least 10x in number compared to the related OMG event earlier this week.

I’ve only had a brief wander around the exhibition hall here; some of the ontologists involved in “government projects” mentioned that good use was being madein those projects of TIBCO BusinessEvents CEP technology. I’m not convinced the semantics world is ready to wake up to “events” and “event processing” yet though (with a few honorable exceptions). I guess we’ll find out later

Another interesting TIBCO technology acquisition with potential side effects for high performance event-processing: the Kabira transactional JVM has joined the TIBCO team. This provides a JVM (called Fluency) that has built-in (to quote the developer guide):
• Transactions
• Distribution
• Shared Memory Persistence
• Keys and Queries
• High Availability
• Replication

Transactional integrity is of course orthogonal to event processing: some event processes, especially those that impact multiple services and processes, need a transactional context. In complex event processing, identifying complex events is rarely “transactional” per se, but could depend on some state / include some transactional context. And of course, the business process associated with a complex event may well need to be a transaction…

… does not, at first glance, have any event processing content. This is unsurprising as most IT folk associate “semantics” with ontologies and the “semantic web”, where the latter is a hugely successful initiative (provided you are an academic researcher or a W3C standard). This success cumulated recently with… yet more government funding (a “Web Science” research institute whose unconventional funding by the UK government apparently raised a few eyebrows).

The Semantic Technology Conference has been growing successfully year on year due to the potential of better semantics in information processing, and probably through being one of the better organised commercial conferences. And a convergence (or collision, depending on your point of view) with the commercial world with the semantic community may not be too far off. Consider, for example:

• the promise of RDF (and associated RDFS and SPARQL) tuple stores for agile data handling, which is but a hair’s width from a commercial, internet-based tuple storage technology like TIBCO ActiveSpaces…
• the promise of declarative logics being applied to information processing, which is but a hair’s width (OK, maybe a small path’s width then) from commercial declarative rules engines like TIBCO BusinessEvents…

Although the RDF/ontology/semantics community is mostly avoiding the event / temporal viewpoint, there is a corner of a corner of Sem Tech that will be well represented by the CEP community. A track within the overall conference is on Rules, and one such session here is described as  “Semantic Rules in the Dynamic IT Infrastructure” with short sessions from Sandpiper (known for semantic CEP), TIBCO and the RuleML team (on reactive rules, CEP and Streaming Knowledge), and a panel with Opher Etzion (EPTS Chair) and Jeff Palmer “formerly of Streambase” (who I don’t believe I’ve met…) [see comments].

SemTech takes place on June 21-25 in San Francisco, USA, with the aforementioned Semantic Rules sesion on Thurs 24th at 6pm.

UPDATE: oops looks like the schedule changed / got updated.