Talk:Knowledge-based engineering

Notice: on marking as Ad
I marked this page as Ad because I thought it was too oriented to CATIA. I hope I do not offend anyone with this template :-)

I hope I have more time later to make the improvements I think are necessary... NonDucor (talk) 18:10, 27 January 2008 (UTC)
 * Not offensive at all. I'm an expert on AI and knowledge based systems and this article needs serious work. It's unclear and seems very promotional as is now to me as well. The first four references are IMO worthless. Three of them seem to be to a company's promotional material and probably were questionable when they worked but as of now they just go off to nowhere and the site never responds. I'm going to try again later on the off chance that a server is down somewhere before deleting the references. I think there are some tools for examining link rot. I've never used them but they might be worth checking out. Of the first four references the only one that goes to an actual page for me goes to a very general AI industry group page and as far as I can see has nothing specific on this topic. The first reference that seemed worth keeping was the fifth one which was to an OMG specification that described the topic. I plan to delete the first four references if they still don't work and if the link rot stuff doesn't work. I'm going to delete the one that is generic anyway no matter what but wanted to document here first in case anyone is watching and wants to comment or discuss. --MadScientistX11 (talk) 04:29, 3 July 2014 (UTC)
 * A follow up comment. I posted a question about this at the teahouse and someone mentioned the Wayback Machine. Here is info on how to use that if anyone is interested: https://en.wikipedia.org/wiki/Help:Using_the_Wayback_Machine  and I found at least some of those links on the archived site. Here is the first one: https://web.archive.org/web/20120324223130/http://legacy.coe.org/newsnet/Jun05/knowledge.cfm  This looks like it was IMO a decent reference. I'm still reading and figuring out how to use this info to make a better article and get a more up to date reference but wanted to post this as an FYI in case anyone else is watching this page. I'm doing stuff in the real world so only spend a little spare time editing and researching. I plan to keep working on this but of course if anyone else has more time go for it. cheers. --MadScientistX11 (talk) 13:35, 3 July 2014 (UTC)

Knowledge, again
Might add: and processes related to such. After reviewing the status of AI and seeing the remarkable updates to the page (see Artificial Intelligence), it seemed that KBE has been too invisible. For over a decade, it was successful in supporting major processes (starting with the 777) related to CAD/CAX. That was 30 years ago where techniques getting a lot of attention now were being done. In short, we were handling the intractable in order to get solutions of real problems under heavy constraints. It was fun. Lots of people wanted to be on the team. For myself, I would like to see less emphasis on the 'AI' as an entity and to move the focus to the mathematics being used. With some notion of results. So, KBE was there in the winter and doing something very useful. CATIA was mentioned. Unfortunately, the vendor took ICAD away. Based upon what I read, there is still interest in LISP. Not everything is numeric (though, some interesting discussions are still pending in that regard). This might seem late but is not. About the time of this subsection (2014), the NN folks were starting to gain some leverage with GPUs and some opportunities with the data that exploded after Steve's gift. So, we needed time to hear all of the stories and to assess what was what. Now, knowledge can be brought back in a more full sense. I know Wikipedia is not for new stuff. 30 years ago is not new. jmswtlk (talk) 01:14, 1 December 2021 (UTC)

Notice:
This was my first attempt at describing this topic. One main point is that we to address concerns of both PLM and CAx. The former is a top-down business framework. The latter is bottom-up science and engineering. There is overlap as would be expected. COE has published a related article. (Marking earlier comment, mea culpa, 9 November 2005) jmswtlk (talk) 18:24, 7 July 2014 (UTC)
 * Too, I did not, at first, mention ICAD (fantastic tool) so as to not show bias. ... Also, I see that this was a comment moved from top to after Notice as a response (see 27 January 2008). jmswtlk (talk) 18:24, 7 July 2014 (UTC)
 * General comment. This page was started in 2005, nine years ago (at least, the first page was marked by me - after that, I kept forgetting to log in). Look of page on 17 March 2006 which was right before the COE Meeting in Atlanta, GA. Some with a KBE interest have made changes. Of late, we have new interest which is good. The page does deal with more than a "low priority" subject (it's mostly that the benes are cloaked under proprietary shields). There have not been many comments until lately, but the main page has had over 600 edits (some good, many not). jmswtlk (talk) 18:57, 7 July 2014 (UTC)

ICAD
I am surprised that there is no mention of the ground breaking ICAD system in this history of KBE

Response to ICAD
Thank you for bringing this up. There is an indirect reference (See Also - A KBE System for the Design of Wind Tunnel Models using Reusable Knowledge Components).

This is my first pass through. Notice that I barely mention CATIA; it's mainly because COE has related pages that I could point to.

ICAD has been buried by Dassault. Do a search on ICAD on Google and see what you get?

Yes, ICAD ought to (will) feature heavily in the history. I'll be correcting that omission (skipped over it by intent, in order to get the larger picture out there). Someone please tell me, how ICAD can feature in the future (without divulging proprietary knowledge)?

ICAD is an archetype (so to speak -- more on this later). I put in a specific page for ICAD; this will allow the historic review that you suggest is necessary. (marking an earlier comment, 3 November 2005) jmswtlk (talk) 18:27, 7 July 2014 (UTC)

++++++++++++
ICAD had some fundamental engineering knowledge management menthods of implementing the design intent in an intelligent way, which the current codes still need a long way to go

Response to ++++++++++++
Granted ICAD seemed to have 'something' (see Joel Orr's "Circle Game" in the Oct/Nov COE Newsnet) that exhibited 'power' in quantifiable terms. One reason for the KBE hiatus was that ICAD vanished (the facts concerning this ought to be described further). Since that time (circa 2002), there has been improvements in the Dassault tools. How and when this collection's capabilities can rival what was there with ICAD are open questions. Perhaps, to answer this question, we need to change some of the evaluation rules.

At the same time, though, we must remember that UG embedded Intent. It would be interesting to see how that set of tools has progressed and to what ends.

As a means to organize the appropriate data, I'm going to change 'KBE Theory' to 'KBE Futures/Theory' and use that section. (Note, marked much later, don't why I didn't sign at the time, 13 November 2005). jmswtlk (talk) 18:19, 7 July 2014 (UTC)

Model
A change to the 'See Also' section focused 'Model' on 'Model (abstract)' which I think is too strict. From the context of the 'KBE Futures/Theory' section, one can note the broad domain covered by KBE. This context allows for two major divisions of knowledge: science (which is acquisition) and engineering (all that mankind does - subsuming technology, too). Hence, we need pointers to both. (marking comment from an earlier date, 15 December 2005) jmswtlk (talk) 18:48, 7 July 2014 (UTC)

KBE Future and Theory Section
The section labelled KBE Future and KBE Theory seems fairly incoherent to me. There is a tag to that effect that was here before I started looking at the article. To the extent it makes sense it sounds like philosophical musings and opinions that would be fun to talk over with the author over a beer at a KBE conference but are in no way appropriate for Wikipedia. Since there are no references I'm leaning toward just deleting that entire section. Want to read again and work on other parts of the article first though, speak up if you see anything in that section you think is worth keeping. --MadScientistX11 (talk) 22:14, 3 July 2014 (UTC)
 * As the author, I agree, in part. Having done KBS before KBE, that there was a trajectory seemed obvious. Actually, think of it as musing about 'bots' (not really known at the time) being able to put on the hat of a domain expert (and way before now when AI is considered embedded). But, alas, the imposition of STEM-based mindsets has poisoned the ability to look beyond the equation (unfortunate, in so many ways). So, how ought the look at the future look? jmswtlk (talk) 22:58, 3 July 2014 (UTC)
 * Thanks for the response. BTW, sorry if my first comment seemed harsh, I tend to just be overly blunt some time, I really did mean what I said the ideas sound very interesting, just not sure they are appropriate for here. For what it's worth I think there is amazing potential, I hadn't even thought of this until just now but it's such an obvious fit, between the Semantic Net technologies and KBE. Things like OWL, SPARQL. Protege, etc. Protege alone, the applications for building meaningful ontologies in CAD is I think enormous. I can't say enough good things about Protege, I've used a lot of knowledge-based tools in the past but never seen one that is so good at the analysis level rather than at the implementation level. Bit of a tangent, back to his page: please keep watching, I'm making minor improvements to the refs, the wording, etc. I'm going to do a search and see if I can find anything on Semantic Net and KBE and if I can maybe use that as a reference for the future stuff. Of course if you have other suggestions for good references on that topic please let me know. I know a lot about knowledge based systems in general but haven't had much KBE experience. --MadScientistX11 (talk) 23:26, 3 July 2014 (UTC)
 * Happy to see the interest. jmswtlk (talk) 00:17, 4 July 2014 (UTC)
 * See KBE Futures, below. jmswtlk (talk) 19:17, 7 July 2014 (UTC)

PLM Section
I removed the Main macro at the beginning of this section. I don't think it was used appropriately. If the section were about PLM in general (what the macro linked to) then yes, that would make sense but this section isn't about PLM in general but PLM and KBE, since there is no such article yet it's premature to use that macro. Instead I put a link to the PLM article at the beginning just as a normal link. I also updated the info about the OMG specs, did some minor word smithing, cleaned up one link so it wasn't a raw URL and added another link to the OMG CAD spec. --MadScientistX11 (talk) 00:32, 4 July 2014 (UTC)
 * One other thing just occurred to me as I was reading through the article. It's not clear to me why the OMG stuff was even in this section to begin with. It seems perhaps there might be some confusion of terms (and possibly I'm the one confused). But my understanding of PLM is we are talking about the life cycle of the stuff that gets built. So if we are doing KBE for General Motors we are talking about the product lifecycle management of cars and their component parts. There are CORBA standards for "lifecycle" management but that's a totally different lifecycle that's the lifecycle of an object (the main data structure used in most modern computing software environments) not of the car or whatever product is being made. It seems to me that the CORBA spec stuff belongs more in the section on languages or somewhere else but not in PLM. It's about the underlying formalisms, languages, technologies used to implement KBE and hence not appropriate for the section on PLM. If I'm missing something please let me know otherwise I'll make that change. --MadScientistX11 (talk) 19:01, 4 July 2014 (UTC)
 * Great work, so far. Let me try to find that spec. It would have concerned extensions related to specific requirements for PLM (a lot of which are cloaked under proprietary shields). With these abstracting efforts, application code would be replaced with general calls (now whether this type of lift was/is of use is something that is interesting to discuss - let's say map/territory issues, abound) to then be used generally. In many cases XML was good enough to meet the need. I notice that we have had academic visitors. Perhaps, they can weigh in. My experience was hands-on, commercial, doing things that were of interests to visiting professors. jmswtlk (talk) 21:28, 4 July 2014 (UTC)

KBE Methodology section
Just want to document why I'm deleting what is currently reference 9: "Knowledge Acquisition in Practice: A Step-by-step Guide" by Nick Milton. It's a dead link but I found it on the wayback machine. Here is the archive if anyone is interested: https://web.archive.org/web/20080409230014/http://www.pcpack.co.uk/Book/ However, this reference is just a general book about knowledge acquisition and knowledge engineering. There is IMO no need to use a stale archived reference for that. There are better references on that topic. I'll find one and add it back in soon. --MadScientistX11 (talk) 15:41, 4 July 2014 (UTC)
 * I found a good ref for knowledge engineering: Kendal, S.L.; Creen, M. (2007), An introduction to knowledge engineering and added that in the spot where I removed the stale link. --MadScientistX11 (talk) 16:12, 4 July 2014 (UTC)
 * I agree. Nothing stale is necessary. But, given that I was one of the COE Newsnet editors for KBE, it might be interesting to pull together all of the KBE articles, such as Prasad's. Or, I might do this elsewhere and put an external link here. As, I also want to summarize what happened over a fifteen-year period at Boeing (as in, taking NOR elsewhere). jmswtlk (talk) 21:33, 4 July 2014 (UTC)

Point-in-time example
In the early 1990s, Boeing was getting the 777 program ready to produce which meant solving problems related to PLM (the gamut) and was, at the same time, attempting to accomplish digital definition (parametric modeling was one option, KBE via ICAD/LISP was another). KBE had an impact on the success of this program (from here, point to articles of the time plus a summary of the performance of this aircraft - to wit, fits were much better than ever seen, alternatives were allowed since the computer models could be changed more easily, ...). Briefly, plane definition is driven by performance goals (aero, flutter, etc.) and system (capacity, etc.) and producability (it has to be built and flown); the definition and build and planning are all heavily knowledge-laden problems, mostly of an engineering and mathematical nature. The external definition looks like the plane (but is analyzable). The internal parts, then, adapt to constraints imposed by the external design (geometry, mechanics (engineering of), etc.). ... In the 777, shear ties (that join stringers and frames for the airframe under the skin) were multiple (very large number, mostly unique). As eternal configuration changes propagated in the design world, all parts would then adapt (meaning, solid modeling for shape as well as other representations for analysis). KBE was able to get the define process to the point where a configuration change could be followed by weekend's worth (SUN servers, RISC, and IBM mainframe) of computing that populated the huge model with new parts. These results were checked by the engineers. To get to this state required a whole lot of innovation and serious work including special mathematical modeling to reduce footprints (point to paper on the Multiple Surface Join and Offset project - providing proper representations to the solid modeler) of the digital model of the part as well as to optimize the definition going to subsequent processes (to wit, produce who had, then, to develop the actual part). ... In short, the success of the KBE project led to successive programs using KBE and widening of the scope (eventually touching about everything). As with any computing, managing expectations related to KBE ought to get some attention (again, briefly stated). ... There are other examples, such as Rapid Forging Design, Intelligent Tool System, interior layout design assistants, etc. jmswtlk (talk) 10:28, 6 July 2014 (UTC)
 * I remember the 777. I never worked for Boeing but I was very involved in what the IT people called User Centric design, it was a precursor to what is now Agile Methods, the idea that users should not just "throw the requirements over the wall" to the developers but should be full time participants on the team and from what I recall the design process for the 777 was very ground breaking in that regard. I think having a case study section or just incorporating this (and other) examples in the article is a great idea. jmswtlk, if you (or anyone) has some good references for this case study please make sure to post them. I'll do some research as well, starting with the links in your comment above. --MadScientistX11 (talk) 14:42, 6 July 2014 (UTC)
 * Good work, thank you. I'm going to put pointers here to papers (talks): McGoey's Hitch-hikers Guide to KBE (Aerospace details), Wind tunnel design (good example of ICAD use), Aircraft wing-body design (example of ICAD and the domain specifics), ... jmswtlk (talk) 20:09, 6 July 2014 (UTC)

Found, today, an old paper from the 1994 time frame which relates to wrapping up the 777 design thrust. The paper was given at a conference in the Kansas City area and was sponsored by Kansas State University. I was on the organizing committee. Papers were from academics as well as from practitioners. Despite helping get the AI winter page started, this paper shows that AI techniques were being studied and used for decades after the start. Some of the work was enabled by Lisp being ported to the general workstation (Unix then MS Windows). Practical issues of AI (from the view of a worker bee, essentially). The techniques involved include NN, higher-order mathematics and statistics, as well as the state-of-the-art software tools. I am still looking for a photo of ICAD use. One paper at this conference has drawings that might work of empennage that discusses the pros/cons of KBE. We found more pros than cons. jmswtlk (talk) 02:02, 15 January 2019 (UTC)

Overview
If this were FB, I would do a 'Like' (if that has any meaning here). What do I like? Going back to KEE which then goes back to MYCIN (and eMYCIN). ... Now -- disclosure: I spent over 10 years with KEE (and its offspring - SIMKIT, etc.) on a slew of different LISP Machines (Symbolics, Xerox, LMI, TI, TI-MAC) and even on the IBM mainframe (very interesting). And, I spent time with the follow-on products, such as LiveModel (see IntelliCorp) and IC's UML Modeler. Then, I spent 12+ years with ICAD/LISP. At Boeing, ICAD was adopted since engineers (all of the disciplines) were easily trained and were very productive using the package. Imagine trying to get a software person to understand engineering trade-offs, etc. Aside: all the work in production environments with heavy requirements - at Boeing, 777, 737 several, 7x7, early 787, others. -- This section deserves its own page, at some point. ... Excellent work, MadScientistX11. jmswtlk (talk) 20:26, 6 July 2014 (UTC)
 * Thanks for the Like! Simkit and KEE were amazing. It's funny that in all the years after I've never found a development environment that had the productivity of KEE on a Symbolics machine. I was thinking of trying to work a bit more into the article about where KBE is now and using Intellicorp as an example. My feeling is that what happened to KBE is similar to what happened to expert systems tools like KEE. There was a lot of disillusionment based on over inflated promises (I'm also toying with working in some refs to the Gartner hype cycle) but the ideas never really went away. Many of the applications ended up being redeveloped under the mantle of eCommerce but mostly they were purchased (or just redeveloped from scratch) by big ERP and similar vendors. So for example the way SAP purchased Intellicorp. Assuming we agree on that it would be nice to have some real life examples of other KBE vendors that were acquired. I'm sure IBM must have purchased some. I think Ariba was also purchased by SAP. Ariba was a B2B eCommerce vendor that had sophisticated AI algorithms for trading and negotiating. I guess that's on the border of KBE, I would consider it part of KBE but not sure everyone would agree. Let me know if you have any ideas along those lines. --MadScientistX11 (talk) 21:10, 6 July 2014 (UTC)
 * Looking at this again, there ought to be some changes (for one, separate page -- due to the amount of material to cover --with a summary on the KBE page). Notice, the text jumps from the 80s to the 2000s. There was a lot of progression from the KEE days to Dassault's current CATIA suite. This means that an overview of KBE isn't about those two alternatives, rather we have steps along the way in the software (but, processes and engineering knowledge changed, to boot). McGooey's talk has the progression as seen from Boeing's view. We can use the view of others, to boot. (That said, the whole article looks much better -- Aside: I still like the Futures in the sense that there are deep philosophical issues to address - not just a matter of who makes the most bucks, so to speak ;-). jmswtlk (talk) 18:18, 7 July 2014 (UTC)

KBE Futures
Perhaps, thinking of the "Futures" section might motivate the proper view. The work is being done (proprietary shields keep us from seeing too much). Now, with your edits and improvements, the article looks good, and others might be more inclined to weigh in now that it's less cumbersome. ... This OMG document (KBE Services for PLM) still stands as representative of the viewpoint (albeit, 2006 -- the cloud'd world, notwithstanding, we still have to think of principles of engineering). [Walter's view from working with a massive systems engineering perspective are apropos] ... As for tools, I don't know if there were many that had applications as heavy as we saw with ICAD (the differences ought to be spelled out, perhaps, general tool versus domain specific tool -- now, the Dassault suite is a heavy hitter). Too, there was Pro/Engineer (one motivation for Prasad's article) and others. ... Yes, tools went away because the larger competitor wanted them out of the way (Dassault put ICAD to bed in that way, so to speak). jmswtlk (talk) 21:34, 6 July 2014 (UTC)
 * Yes, the Futures section is one I'm still not certain what if anything to do with. If you get a chance check out my sandbox: User:MadScientistX11/sandbox I have a draft there now of what I was thinking for the Futures section but I'm not satisfied with it and can't really think of much I can add. I think everything I say there is true but it's really vague and I don't have any really good references. I was toying with just deleting the Futures section, I'm not sure we have much to say there right now and the article has plenty of sections as is and some futuristic stuff is covered in other sections, where I mention Semantic Web. --MadScientistX11 (talk) 04:28, 7 July 2014 (UTC)
 * I'll do that this week. For starters, we can use McGoey's take (see slide 22: The Future of KBE). Also, he did an overview of KBE plus a history (see slide 12: KBE Usage in the industry - starts in the '80s). Plus, Paul has a lot about software/process/systems engineering. ... jmswtlk (talk) 10:09, 7 July 2014 (UTC)
 * User:MadScientistX11/sandbox Early thought. Expectation management is only aspect. In terms of process improvement and cost reduction, KBE has played more than a minor role (well documented savings, observable quality improvement, maturation of models, resolution of map/territory issues, ...). The thing is that as capabilities progress more is wanted (from the management perspective without dampening) such that unrealistic notions can ensue (water runs uphill sort of thing). So, the bit from Gartner might be a separate section (perhaps, a later general view about progress and expectations). The other sections ought to use material that is based upon specific experiences from those who wrote of their work, like McGooey, Wilson, and more. Then, there needs to be some mention of the deeper issues -- see above, Overview, brief comment about philosophical issues - quasi-empiricism, for one). jmswtlk (talk) 18:18, 7 July 2014 (UTC)
 * jmswtlk, I mostly agree, except not on the "deeper issues". IMO the stuff about Functional Programming or the Entscheidungsproblem really don't belong in this article. Just to be clear I love that stuff. I actually really found the Entscheidungsproblem article fascinating but I just don't think it's appropriate for this article, it belongs elsewhere. I'll use an analogy to Expert Systems because I have a lot more knowledge of the business world there but I think this article is very much like the Expert Systems article in terms of what makes sense and what doesn't. In Expert systems there are also all kind of interesting and mostly unsolved problems: the frame problem, truth maintenance, common sense reasoning, the role of natural language, etc. Actually as I write I'm even realizing that all that stuff could be added to the expert systems article and it might be useful but it would be a new section, on theoretical issues. So I guess I would change my opinion in mid comment, if we keep the stuff like Entscheidungsproblem it should be separate from a KBE futures section and in it's own KBE Theoretical Issues section. Having said that I honestly don't think the theoretical stuff that is there right now makes a lot of sense anyway. I'm skeptical that the Entscheidungsproblem is really that critical to KBE but as I don't know as much about KBE as expert systems I can't say definitively and I'll leave it to people like you who have more experience to make that judgement. --MadScientistX11 (talk) 18:51, 7 July 2014 (UTC)
 * MadScientistX11 Looking forward to seeing the page after your edits. Please, though, stop by the first cut of that section (see 17 March 2006 to see how simple it was). Now, whether the changes, through time, have improved or degraded the thing would be an interesting study. I'll stay out of the way until you are done. The KBE community appreciates your effort. jmswtlk (talk) 19:15, 7 July 2014 (UTC)
 * jmswtlk Thanks. I was going to just leave the futures section but I did a search on KBE and SAP and was happy to find that my hunch seemed right on the money. There were LOTS of papers and web sites about how SAP's manufacturing systems were now knowledge-based and the term "KBE" was still used quite a bit. I just picked the first one I found as a reference but the fact that there were so many made me feel that I had enough foundation to say what I wanted to say. I left off the very theoretical stuff. My suggestion is if you want to bring some of it back create a new section called KBE Theory. IMO the Future stuff should be more focused on the practical application of the technology in the real world where as theoretical stuff is still important but should be off on it's own. I think I'm going back to doing non-Wiki stuff for a while, stuff that pays the bills :) but I wanted to finish that one bit so we could get rid of that tag about incoherence. Cheers! --MadScientistX11 (talk) 19:32, 7 July 2014 (UTC)

---

It turns out that KBE is a wonderful model to help work issues related to computational thinking. See the page on Operation Definition (In computing). After all, experiments are very much results of work done to engineer the framework. jmswtlk (talk) 14:51, 2 January 2017 (UTC)

External links modified
Hello fellow Wikipedians,

I have just modified 2 one external links on Knowledge-based engineering. Please take a moment to review my edit. If you have any questions, or need the bot to ignore the links, or the page altogether, please visit this simple FaQ for additional information. I made the following changes:
 * Corrected formatting/usage for http://legacy.coe.org/newsnet/Oct05/index.cfm
 * Corrected formatting/usage for http://web1.eng.coventry.ac.uk/moka/default.htm

When you have finished reviewing my changes, please set the checked parameter below to true or failed to let others know (documentation at ).

Cheers.—cyberbot II  Talk to my owner :Online 10:38, 2 May 2016 (UTC)

External links modified
Hello fellow Wikipedians,

I have just modified one external link on Knowledge-based engineering. Please take a moment to review my edit. If you have any questions, or need the bot to ignore the links, or the page altogether, please visit this simple FaQ for additional information. I made the following changes:
 * Added archive https://web.archive.org/web/20130424071228/http://www.cs.umd.edu/~golbeck/LBSC690/SemanticWeb.html to http://www.cs.umd.edu/~golbeck/LBSC690/SemanticWeb.html

When you have finished reviewing my changes, you may follow the instructions on the template below to fix any issues with the URLs.

Cheers.— InternetArchiveBot  (Report bug) 05:27, 7 May 2017 (UTC)

Real world examples
Under the context of ICAD (Talk), there have been a couple of examples of KBE written. Now, the emphasis might have been on how KBE works. Given the proprietary nature of this type of work, it is hard to find examples. So, I also started a new reference list on that Talk page.

This stuff really belongs here. If you want to contribute, let's work this. I used two examples from Boeing: RFD and monocoque work.jmswtlk (talk) 02:09, 15 January 2019 (UTC) (forgot to sign)

See this paper (Practical issues of AI) from a 1994 conference. The TOC of the proceedings follows the paper. There are lots of examples in the proceedings. I have the physical book. Also, see Talk:Knowledge-based engineering (edit this day). jmswtlk (talk) 02:09, 15 January 2019 (UTC)