I could not agree with you more!

Current CEP engines have very implemented capabilties; and can best be referred to as SEP (simple event processing).

Warm Regards,

Greg

The notion of “object” is really wrong. Developers need to understand that concepts are not necessarily objects. Object-orientation is very implementation-oriented. For example, money or time are not objects, but they are concepts and their semantics is important. Replacing uses of “object” with “concept” is a good first step towards clear semantics and longer lasting designs.

I have a few pet peaves about RDF, everybody does! The fact is that its uniformity is both powerful but inconvenient. I wish it could represent quantities without a ton of links and that links could reference other links, but it’s good enough and the train has left the station.

I am not too caught up on the specifications. The arguments are and should be on what common concepts are and mean. With meaningful standard-independent consensus on such definitions, any standard would be good enough for many practical applications.

For Mark: why not download Nexpert, with its docs? (Google is your friend!) This is primarily a bwd chaining engine where bwd chaining was just one agenda option (customizable per domain class, and inheritable) for evaluating boolean terms (if I recall).

Although while it shows the “needs” mechanism, it doesn’t say how OPSJ provides that data, just how it requests it.

James Owen often raves about Neuron Data being the only Object Oriented fully opportunistic forward and backward chaining system. Anyone got any old manuals on NExpert, I’d be interested to see how they did things. Same for ART

I’m curious to hear what you think OWL. I’ve studied it a bit over the last few years and I am left with several important questions.

1. since very few tools use OWL to model entities and/or domain models, would OWL be better, if it extended XML Schema instead. Atleast that way, most of the modeling tools out there could work with it.

2. Should the relationships between instances of entities be captured as rules? More specifically, OWL has sameAs, and differentFrom. In my mind, that should be captured with rules using Knowledge base techniques.

3. should OWL include a rule language? I ask this because people need a way to reason over the ontology.

4. although many consider building ontologies a different activity than object modeling, the average developer thinks in terms of object modeling. For an ontology to be practical, effective and widely used, doesn’t it have to address object modeling?

5. OWL is often associated with and used with RDF. I find the triplet design of RDF flawed from a modeling and reasoning perspective. If you could start over and design RDF, what would you have done differently?

I like the idea of using Ontologies to enhance BRMS, BPM and CEP, but none of the current public specifications seem to be appropriate for real world use.

peter

In the interest of fairness and accuracy while I’ve spent eleven years in and/or working with Inference/Brightware/MindBox I currently don’t have a dog in this hunt. I’m an independent contractor and my mantra is “Is this a knowledge driven problem? If the answer is yes what is the appropriate representation schema? What is the the appropriate tool given budget and skill set availability…one size does not fit all.”