Saw your comment on the ebizQ Forum, and would love to have you join. Please email me if interested.

Best,

Peter

“And why weren’t other algorithms like Bayesian networks used more in CEP (good question…).”

–> Actually we do use other AI-algorithms, e.g. in the case of the reference model for non-deterministic approaches like Smart respectively edBPM-based FraudManagement. You showed it in your blog-entry about the 5th EPTS-symposium in Trento. We have some ongoing PhD-theses in the domains of retail, banking, insurance and nowadays gaming where fraud seems to be a big problem in the meantime. We will discuss such approaches in our next edBPM expert meeting in Dec 7-8 in Regensburg, you can see it from the draft-agenda, second day.
The theses use CEP and its dífferent EPL-approaches (SQL-like, rule-based, special “prorietary” langauges etc.) for the deterministic filtering of known fraudulent event patterns and Bayesian belief networks, neural networks etc. in an upper tier for the recognition of unkown, suspicious event patterns.

Best, Rainer