User:Howard C. Berkowitz/Strong Articles

From Citizendium, the Citizens' Compendium
Jump to: navigation, search

Orphaned articles, isolated articles, or walled gardens of articles, are problems because it is difficult or impossible to reach through following a logical set of wikilinks. Readers can find the material only if they enter just the right search string, which does not let them take best advantage of Citizendium's knowledge navigation. It is Citizendium's policy that its articles should be strong, which have at least three strong links pointing to them.

Citizendium's metadata template now has an "underlinked" question. By definition, a strong article is "underlinked=no".

A strong link is a wikilink that is:

  • in the body of an article
  • in a Related Articles page
  • in explicit indexing pages
    • a redirect to a subsection, that redirect having an associated definition
    • a common prefix, such as AN-

An article is orphaned when fewer than three strong links point to it. An article is isolated when it cannot be reached through a series of strong links from the main page, a workgroup page, or a Core Article. Walled gardens are a set of articles that have strong links among one another, but all articles within the set are isolated.

Draft proposal

Brief descriptive title: Encouraging strong navigational linkage of articles


All authors are encouraged to make articles, at creation time, strongly linked into the navigation system. Articles may not progress beyond Developed if they are not so linked.

Name and date of original proposer

Howard C. Berkowitz 03:28, 10 December 2008 (UTC)

Username of driver

Howard C. Berkowitz if there's no conflict with Editorial Council secretary

Next step

  1. Agree on definition of "strong" (the "lemma" or whatever is the one remaining weak spot)
  2. Agree that "strength is good", and what exceptions apply.

Target date for next step

Almost immediately, immediately start discussion if lemma/permanent stub isn't a showstopper.


Definition of "strong" (see below)

General discussion

For authors of new articles

When writing an article, consider to what it can link, and what should point to it. If the author cannot come up with three potential strong links to it, perhaps that means that some more general articles need to be written first, to establish context.

Those strong links need to be created, so part of creating a new article is editing three or more pages so they link to the new page.

For workgroup editors

Editors, perhaps assisted by some future automated tool, need to stay aware of orphaned, isolated, or walled material. Creating strong links to an article does not constitute substantial editing of the article; an editor can still nominate that article for Approval if the editor made changes to the article itself, or to articles linking to it, only to prevent orphaning.

For Citizens, and perhaps Indexers

In the process of reading articles, a Citizen may come across an article that is not strongly linked. If it is possible to create links that make sense, doing so on the spot is encouraged. Certainly, a lot of articles now in place don't meet the strong linking goal, because it wasn't a stated goal when the article was written.

Role of the techniques in work planning

Hypothetically, if one did not know a specific case or policy, such as a topical matter of an Israeli detention of a Palestinian, or a Russian detention of a Chechen, one might be able to find the topic by starting at a "law" or "military" page, or perhaps name of country, and work one's way down to a page that gets to the subject of interest.

Remember that a Related Articles page almost certainly will start out Strong, and is a logical anchor point for more specific articles. When one has a specific article and is not sure how to put it into the system, a basic method may to start at a Workgroup or Core Article page, and then work down, perhaps creating Related Articles pages, even without having an associated full articles, until there are reasonable links to the specific topic.

Underlinking and strength

Right now, the article metadata asks if the article is "underlinked", which I understand to be a subjective call of the author. It is fair to say, I believe, that if an article is "strong", that automatically "underlinked=no" is TRUE. While I recognize it would take programming, code could, as soon as a newly created article is saved, count links to it and possibly set "underlinked".

Would the proposal be better if I emphasized that "strong" is the antithesis of "underlinked"? Do people really use "underlinked" for any particular purpose? We don't, as far as I know, have an "underlinked article" special page, but, using this definition, one could be created. Especially if it was organized by workgroup/discipline, that would be an ideal opportunity for a subject matter expert/editor to look at new articles and perhaps see linkages that a less experienced author might not. Howard C. Berkowitz 19:12, 11 December 2008 (UTC)

Open issues

I think the key open issue is something that doesn't have a great name. Permanent stubs has been suggested. Lemma may give the flavor. It's a class of article that is legitimately isolated, because the only reason it exists is to have a "linkable definition" to be used within a small set of articles, and is implausible to be an external search term.. Such an article would be isolated but not orphaned.

Whether or not it is the best example, consider Constituent country‎. Is this an example of something that would only be used in the Walled Garden of Great Britain, which does sound like a promising book title? I've never seen the term used outside that context; would it be unlikely, then, that it would be a search engine argument and justify a cluster? In our terms, does it need to be able to link up to politics or geography? Does it need a constituent country/definition? How should it appear in an R-template, especially in Related Articles pages for Scotland, Great Britain, Wales, etc.? Howard C. Berkowitz 19:17, 6 December 2008 (UTC)

At present the working example for your "lemma" is Constituent country‎ and {{R|Constituent country‎}}, gives:
Obviously such a short article can exist as a definition only too. Would that not be the best route here? This then raise the old issue of how to deal with the inevitable red link. A redirect to the definition is one route. Possibly we could have the subpages template define all articles that have a definition only but no metadata in a "Lemma" category? (might need to think about this some more) Chris Day 19:27, 6 December 2008 (UTC)
Good questions that get me thinking. I'd be hesitant to make the assumption that all definition-only pages are properly lemmas. This, of course, begs the question, "what is a lemma", and I think I have a better idea of what I really mean by a lemma. In addition, another idea is developing, the anchor.
Are anchors and lemmas the same thing? Maybe. They are similar in that they exist to be the destination of links from articles. They are different in that multiple articles have lemmas as more-detailed "child" information to which they refer in a consistent way and that anchors are common "parents" do not have much unique content. It should be noted that the children of an anchor are, however, legitimate.


A lemma is a term or concept that is common to some group of articles, but would be unlikely to be a topic for which a user would search. It is analogous to a utility function in software, in that it avoids rewriting, in a possibly inconsistent way, a specific set of actions.


An anchor is an article that might have a small amount of description, but is there mostly to be sure that the detailed articles have a common parent topic. Example: Provinces of Vietnam. The closest current equivalent is "list of foo". "List of" vaguely bothers me, as it would seem that the "of", more often than not, should be in the Subtopics section of a Related Articles page of a Lemma. Example: AN-. For the computer science type, an anchor is a trie, as opposed to a tree. In a tree, non-terminal nodes only point to other nodes. In a trie, a non-terminal node also can contain application information.
Redlink is indeed the first problem, as well as the related italic blue link produced by R-templates. The second is that it not be picked up as "uncategorized" so some helpful person puts subpages on it.
The third is that while it can be a short article or a definition-only, it has aspects of both. One of my wild thoughts is that with some kind of flag, it only needs to be created as one, and the text automatically transcludes to the other. Howard C. Berkowitz 19:33, 6 December 2008 (UTC)
This is early days but i just ran with your idea and made a template that can exist on the article but transclude the definition only subpage. I could rewrite the {{r}} template to get rid of the italics in this scenario, as well as rewriting the subpages template such that the {{lemma}} template is automatically placed on an article if a defintion subpage exists but there is no metadata. Therefore, the only template that would be needed on the article would be the subpages template. Let me know what other features are desirable. Chris Day 20:27, 6 December 2008 (UTC)
Much of this is sounding more like a Zen koan. Let me see if I understand. At the time of creating Foo/Definition, in edit mode, something indicating "lemma" has to be included. Once that is done, the subpages template (is that the right word? Template? Page) gets created automatically, but the Foo (article) gets created automagically and transcludes the definition? In other words, all the author actually creates is the Definition? Howard C. Berkowitz 20:37, 6 December 2008 (UTC)

The author would presumably create the definition only subpage first, probably in response to a request from using the term with the R template. Then there would be the red link in the r template. To remove the red link the author could choose to create the article with nothing more than the subpages template. In that scenario the subpages template would transclude the definition in lieu of a real article and categorise it as a Lemma article, or whatever. At a later date, such an article could be expanded to be a real article with metadata. Chris Day 20:55, 6 December 2008 (UTC)

OK, let me recapitulate. Create Definition, with {{subpages}}, which would cause the Metadata template to show up when the Definition is saved. Fill in and complete Metadata. Where does the "lemma" coding get inserted? Howard C. Berkowitz 21:04, 6 December 2008 (UTC)
No, there would be no Metadata. The definition would be a definition only class. If the writer thought there was no need to have a red link then the article could be populated with a subpages template too (to get rid of the red link). The subpages template would detect a definition only subpage but as there is no metadata it would thus classify the article in the Lemma category and transclude the content from the defintion subpage. Chris Day 21:38, 6 December 2008 (UTC)

Note the advantage of this system is that good citzens who place the subpages template on your lemma article would not be undoing your good work. In theory the subpages template should be used on all pages and then manage the appropriate headings and categorys automatically. Authors should not have to make these decisions. All in all, this will help reduce the complexity for the writer, although it will still exist from the perspective of the inner working of the subpages template. Chris Day 21:42, 6 December 2008 (UTC)

This edit will make you laugh. So our xample is already being modified. And this is a good thing.
Interesting thought here. If that had been the {{subpages}} template on the page that was directing the transclusion from the definition only page, the writer might have left it in place, unlike the way they removed the {{Lemma}} template. If that were the case then the definition would still appear at the top of the page AND the additional inofrmation could co-exist with it.
At some point in the future, as the article expands beyond a stub, the metadata template could be created and workgroup categories, instead of the lemma category, would be used. At that point we would have evolved from Lemma article to a Cluster. One problem with the Lemma strategy, as outlined above, is that the articles in the lemma category would never be able to have workgroup categories too. Chris Day 22:15, 6 December 2008 (UTC)
Well, I've never really looked at tl, so take these comments in that context of ignorance.
I'm personally not strongly convinced that having, or not having, workgroups, is a critical issue now, certainly when there is little or not workgroup-based collaboration in a number of areas.
It may be my personal bias, but I find that filling out essentially repetitive information in a metadata page, manually creating a talk page, etc., are annoying enough that I don't do them unless I have a specific reason, mostly to get them to show up in live articles. If the process were somewhat more automatic, not forcing the multiple separate edits and saves, I might feel better about it all. Frankly, I am mystified on why talk pages aren't created automatically. Howard C. Berkowitz 22:21, 6 December 2008 (UTC)
If we had programmers we could have the whole process automated. Ideally a few drop down menu options would be all that is required to assign a workgroup. Chris Day 22:24, 6 December 2008 (UTC)

How to create a strong article

Once you have an idea for an article X, there are several things that can immediately help organize your thinking.

  1. Write a X/Related Articles page that, in particular, shows the parent topics and perhaps related topics.
  2. Write an X/Definition
  3. Write the Introduction to X, following the basic conditions of intros/ledes.
  4. Write the Metadata, including Workgroups or asking that Editors select them.

Remember that a parent topic can be the core article of a workgroup. If you can't think of a workgroup to which the article belongs, definitely bring this up in the Forums. It may be that you have not searched sufficiently to find a logical parent(s), since few topics of knowledge are completely independent.

Metadata is less optional now, because the Draft vs. Approval status requires metadata to be displayed.