Talk:Metamodeling

From Wikipedia, the free encyclopedia

This article is related to Wikipedia:WikiProject Method engineering.
	Please note that the use of Wikipedia to host this project has been questioned. Please read this discussion and, if you wish, contribute your thoughts there.

Contents 1 Proposed merge(s) 2 Orthography 3 Confusion 4 Reproposing Merge Feb. 2007

[edit] Proposed merge(s)

I agree with the recommendation to merge (purge?) the article on so-called meta-modeling technique with either the metamodel or meta-modeling pages. (Those pages should also be reconciled, but that is a different matter.) The meta-modeling technique article is haphazard and poorly written as an elaboration of metamodeling; it is basically a scantily researched, intro-level academic term paper (not even really original research).

Any discussion of metamodeling to be credible at this time needs to address the [OMG's Meta-Object Facility.] The articles on metamodels and meta-modeling are headed in the right direction. Charles T. Betz 04:46, 21 April 2006 (UTC)

I also support the proposed merge(s). --Allan McInnes (talk) 03:09, 24 April 2006 (UTC)

I strongly support also the proposed merge. We need soon to have a consolidated organization for these central issues of metamodel and metamodeling. How to proceed? MDE 20:33, 13 May 2006 (UTC)

How to proceed? Well, the proposal to merge has been up for a while now, and no one's objected. So I guess we need to start moving any useful content from meta-modeling technique to meta-modeling, while filtering out anything that contains excessive, unencyclopaedic detail. --Allan McInnes (talk) 20:42, 13 May 2006 (UTC)

Most of the material has now been transported. The article meta-modeling technique now only contains only very specific material that does not correspond anymore to general topics. Ready for the next step? MDE 08:34, 27 May 2006 (UTC)

I've had a chat with the Method ENgineering people at Utrecht University. The "meta-modeling technique" is a meta-modeling technique developed there together with another university (in Japan?). It doesn't have a name and is therefore simply refered to "meta-modeling technique". Could be merged in here as an more concrete example, as long as it is made clear that it is only one possible way to do meta-modeling. —Ruud 13:34, 27 May 2006 (UTC)

[edit] Orthography

The article presently states:

The recommended standard orthography is metamodeling

By whom is this orthography recommended? --Allan McInnes (talk) 21:39, 13 May 2006 (UTC)

The trend is present papers is to use metamodel and not meta-model. This is not a critical issue but I believe we should also start challenging all usages of meta-model or meta-modeling in wikipedia to give a justification of thhis spelling. MDE 20:07, 16 May 2006 (UTC)

Do you have a citation for this trend, or is this simply your own observation? If the latter then it constitutes "original research", which is not permitted under Wikipedia policy.

As an aside: I have no opposition to establishing a standard orthography for use here on Wikipedia. What I object to is inserting assertions about standard orthography into the article itself, if no such standard orthography verifiably exists. If you want to propose (on a talk page) that all Wikipedia articles should use the spelling "metamodel" I will happily support you. If such a proposal is adopted, I suggest that the first action be to rename this article to be consistent with the orthography you wish to use. --Allan McInnes (talk) 20:37, 16 May 2006 (UTC)

About the recomended spelling for metamodel, I added as an argument that: "This is consistent with classical usage like Metaphysics, Metaprogramming, Metacognition, Metamathematics, Metalogic, Metalanguage, etc. " This has been deleted without justification. However I still do believe this was a neutral point of view and a good argument. MDE 20:11, 16 May 2006 (UTC)

I removed that statement because it did seem like an "argument", rather than a citable fact, which would again make it fall under WP:NOR. If you have a citable source that says something to the effect that "we should use the spelling 'metamodeling' to be consistent with Metaphysics, Metaprogramming, Metacognition, Metamathematics, Metalogic, Metalanguage, etc." then I'm quite happy to include the statement. --Allan McInnes (talk) 20:37, 16 May 2006 (UTC)

It seems here that the strong defenders of the hyphenated spelling (as in meta-model) are going to ask the other camp (non-hyphenated as in metamodel) to prove that they are right. This does not seem fair to an outside observer. The arguments proposed showed that the natural spelling of meta* does not use hyphenation. So the propagandists of meta-model should have the charge of the proof. 58.32.133.94 00:21, 23 May 2006 (UTC)

You misunderstand. I am not a defender of "meta-model". I frankly don't care which version of the spelling gets used. I care about two things:

1. Consistency — whichever spelling we settle on, let's make sure it gets used for consistently (which means renaming articles if the consensus if "metamodel" rather than "meta-model");
2. Verifiability — assertions within the article about which spelling is preferred must be backed up by citation, not by argument (regardless of how compelling the argument is). See Wikipedia's policy on "original research".

I am quite prepared to concede that MDE's argument for the use of "metamodel" is a compelling one, and that we should therefore use that spelling in preference to the hyphenated version when writing articles. But that does not mean that I support making a statement of that preference part of the article itself.

--Allan McInnes (talk) 01:12, 23 May 2006 (UTC)

So seems that there is finally a consensus here. First we should delete MDE's argument from inside the article. Then we should do some cleaning since the non-hyphenated version is used much more frequently in many other contexts. Should we? Or should we still wait that the defenders of the hyphenated versions (meta-model, meta-modeling, etc.) present their arguments if they have some?—The preceding unsigned comment was added by 62.23.3.58 (talk • contribs) .

--I have removed MDE's explicit argument to take into account Allan McInnes's suggestion. Hope this is OK to you. If not please revert.—The preceding unsigned comment was added by 62.23.3.58 (talk • contribs) .

Any new initiative to complete the cleaning up of this article? 82.67.133.100 17:29, 5 June 2006 (UTC)

[edit] Confusion

I'm having some trouble understanding what metamodeling is from this entry. I try to relate it to what I know and I remain confused.

The opening sentence states "metamodeling is the construction of a collection of "concepts" (things, terms, etc.) within a certain domain". That just sounds like creating a data model. So why isn't that "modeling" instead of "metamodeling"?

I want to build up a lending library application. I have books, CDs, and DVDs and I lend them out to people and organizations. Using pure data modeling, I create Book, CD, DVD, People, and Organization types. I normalize the model and factor out Lendable and Borrower as base types. Those 7 types are my "model". The books, people, etc in my data store are my "data". (I state those last two things explicitly to make sure we have the same understanding of "data" and "model".)

In new-school terminology, I'm making an extendable lending application. I create the base types Lendable and Borrower. I create a mechanism so that people using my application can create derived types to model their situation. In this world, I have created a metamodel that allows people to create a data model. Maybe they make one for books, CDs, and DVDs or maybe they make it for power tools and model trains.

Is that correct? Are the types Lendable and Borrower the metamodel? Should Lendable and Borrower be abstract (non-instantiable) types? If they aren't abstract, are they part of the metamodel and the data model? Is the code (or definition language) I write that lets people define their own types (Book, PowerTool) the metamodel?

Is a general purpose data modeling tool or language a metamodel? UML? An RDBMS language like SQL? XSD?

Perhaps the answer is I should go get a basic text book on metamodeling and read it so I can contribute constructively instead of just asking questions. Ishi Gustaedr 21:53, 2 August 2006 (UTC)

Like you, I came here trying to understand the difference between modeling and meta-modeling. I agree that I am confused after reading this. I'm just going to disagree with one thing you say: you think you should read a basic textbook so you can contribute constructively. I, on the other hand, think this should _be_, in a sense, the basic textbook. So this is where you should come _before_ you go to the treeware text!

My original confusion between modeling (henceforth M) and meta-modeling (henceforth MM) was because I had been working with a group for several years doing what we all called M. We were modeling dictionaries (lexicons, actually, but the diff is irrelevant for this discussion). Recently I met someone who is doing exactly the same thing, but he calls it MM.

Here's what I finally decided must be the difference (before I came to this article and got confused again). Previously, we were modeling yet-to-be-built dictionaries. Based on the model, we built software to allow linguists (working in mostly undescribed, and often previously unwritten, languages) to build a database of the words they collected. Eventually, they might create print dictionaries based on the database. But the key is, the _real_ dictionary was the database; any given print dictionary might leave out words that the linguist had collected, not to mention having a different format. So what we were modeling was the "real" dictionaries, and any print dictionaries that might result were derivative.

The person I met, on the other hand, was building databases of print dictionaries. Any particular database was itself a model of a print dictionary--and might leave out things (like page numbers, cover pages etc.) that the "real" dictionary had. So the framework he was building that would describe the database was a meta-model, because it was a model of the database model.

Does this make any sense as the diff between models and meta-models? Mcswell 01:50, 18 January 2007 (UTC)

Agree the article is in need of elaboration. My understanding comes from reading the Meta Object Facility version 1.x introduction, David Hay's recent book on metadata, and struggling through some of Jean Bezevin's work. Don't have time to work any of this into the article which is too brief and a bit on the academic side. It's an important enough topic to merit further treatment, including examples. The issue of relativity needs to be discussed in depth (one person's data is another person's metadata - it is perspective-dependent). Good luck and hope this helps. Charles T. Betz 03:31, 18 January 2007 (UTC)

Thanks, and I look forward to the discussion of relativity (and examples always help, too)! Mcswell 13:45, 21 January 2007 (UTC)

[edit] Reproposing Merge Feb. 2007

Ruud Koot has proposed that these this and Meta-modeling technique be merged. I have updated the date tag at Meta-modeling technique and placed a merge tag on this page. I take no position on whether or not these should be merged. --Selket ^Talk 00:34, 28 February 2007 (UTC)