Talk:Knowledge engineering

Untitled
is it a subset of the Zachmann framwork?


 * What is the Zachmann framework?

There are several statements in this article that are assumptive and erroneous. They are:

(1) Knowledge engineering for rules based engines offered by Fair Issac do not require programming.

(2) Knowledge engineering as taught at the University of California, Irvine: "Creating Systems That Know" by Richard L. Ballard Ph.D. does not require programming, in fact, it replaces programming.

(3) Artificial Intelligence, logic and mathematic systems are contrary to "theory-based" knowledge acquisition and management systems since logic, self-consistency and adherence to true/false outcomes automatically eliminates a machines capacity to reason with uncertainty. For this reason, any system that is based on these paradigms can only process and make artificial assumptions about information.

(4) Another view of Knowledge Engineering is the theory-based semantic view, which states that knowledge is language independent and that semantic meaning is precisely captured as a "language of thought" created by concept relationships and are based on well-justified theory such as scientific and axiologic theory.

Just some thoughts.

Dxthom

Why is electricity not a fuel?81.77.191.145 19:37, 28 November 2006 (UTC)

Intro should be clearer
The introduction seems very convoluted to me. I would start by just saying something like "Knowledge engineering is the process used to develop an expert system". That's pretty much the definition. We can get into the details later but I think the intro should be reworked to be say it in simple English first. --MadScientistX11 (talk) 23:05, 3 July 2014 (UTC)
 * ✅ --MadScientistX11 (talk) 17:04, 8 July 2014 (UTC)

Why I tagged the Trends Section
I've done a lot of expert systems work in my career. The current Trends section in this article is incoherent and has wrong information. For example, comparing KADS and Protege is ridiculous. It's like comparing RUP to a UML editor. KADS is a methodology Protege is a modelling tool. And saying Protege is "influenced by the concept of Ontology" is confusing. Protege is an ontology editor, it's not "influenced by" the idea of ontologies, it just IS a tool to create and edit ontologies and in any case Protege has little to do with expert systems anyway. It's for Semantic Web systems and for ontology editing but not at all limited or even focused on expert systems. That whole section seems to be based on one paper, a paper I've never heard of and one that is not at all representative of where the AI community is now on expert systems. I think the whole section as is now should be wiped and we should start from scratch. --MadScientistX11 (talk) 12:47, 8 July 2014 (UTC)
 * Now that I've read through the whole article more carefully I want to amend this a bit. The whole article is incoherent. It doesn't address the main issues: what is an expert system? what are the various processes used to build one? what kind of stakeholders are involved? What are examples of knowledge engineering methods (e.g. KADS). These are the kinds of questions I would expect to see addressed here and they aren't at all. I think the whole article needs to be rewritten. --MadScientistX11 (talk) 14:19, 8 July 2014 (UTC)
 * ✅I just rewrote the article. Still needs lots of work but this at last communicates the basic concept --MadScientistX11 (talk) 17:03, 8 July 2014 (UTC)

Knowledge Engineering with Bayesian Networks, Kevin B. Korb and Ann E.Nicholson
I believe this article would be improved with a reference to the Knowledge Engineering work described in the book Bayesian Artificial Intelligence. However, I'm a reasonably new wikipedia contributor so before I add anything to the article, I'd like someone to confirm this work is significant and should be referenced here. If there is agreement, I'd also like a suggestion on the best words to use for the reference.