Page MenuHomePhabricator

Identify which are the essential article types for English Wikipedia to be supported with Article Guidance
Open, Needs TriagePublic

Description

As part of the work on Article Guidance (T396029), a new workflow for article creation is proposed and evaluated in an experiment with a set of pilot wikis (T414448). If the experiment (T414536) show positive impact, we may want to scale the initiative, and English Wikipedia seems like a community that can benefit from it, given the volume of articles created and the challenges reported by reviewers.

Given that English Wikipedia has the largest coverage of articles, we may want to look into more detail on which are the common types of articles created there. This ticket proposes to identify which are the essential article types for English Wikipedia to be supported with Article Guidance. To learn about this, we can incorporate input from experienced reviewers as well as data analysis on articles created.

Event Timeline

This is the easy part. Phab is possibly not the ideal venue to address this. All that needs to be done is to collaborate with the people who actually do the daily work of seeing the entire inflow of new content.
Rather than trying to sientifically parse new articles into types and then present the new user with walls of text or confusing drop-down choices (UX and UI psychology), we have whittled this down to 10 -12 basic article types (given that the vast majority of articles on the en.Wiki nowadays are biographies.) The effort should be to avoid thinking up a long list of article types and then spend months debating what the page templates should look like. The recommendation should be to encourage feedback and ideas from the Wikipedians who have a vast resource of institutional memory and anecdotal evidence: the most senior New Page Reviewers from the user right I created in 2016.

'To learn about this, we can incorporate input from experienced reviewers as well as data analysis on articles created.'
Apologies - didn't see that. As former coord of NPP for nearly a decade, don't hesitate to ask for suggestions how to best obtain the feedback and from whom. on en.Wiki. Below is a list of the suggestions for the main types of articles based on a rough 24 hour sample (~500 articles) on 19 March 2025. On that day 28% were biographies (BLP and other Bios):

  1. BLP (Biographies of Living Persons)
  2. Bio (Other biograhies)
  3. Settlements
  4. Sport events (all matches, Olympics, national and other competitions)
  5. Genera (Species)
  6. Election results
  7. Military (History, wars, weapons, armies, navies, air forces,etc)
  8. Movies, TV shows
  9. Video games
  10. Organisations (Companies, Schools, Political Parties, Charities, etc)
  11. Publications (books, journals, periodicals, etc)
  12. Music (bands, orchestras, compositions, concerts, tours, but not musician biographies)

Thanks a lot, @Kudpung . That's already very useful information.

Given the prominence of biographies, to which extent do you think may be helpful to provide support to more specific sub-types of biographies. Based on input form other wikis, apart from a general Person outline, the following were identified as frequently created (or particularly problematic) types of biographies:

Would those be also relevant for the case of English Wikipedia, based on your experience? Should we consider other specific types of biographies instead based on the specific needs of English Wikipedia?

Thanks!!

@Pginer-WMF, yes, I do think subsets for bios could be considered but I would guard against creating too many types of article templates at least until some have been tried out for impact and run past UX and UI designers from the community's veterans first.
The effort is to help the new, new page creators to create their articles rather than doing it for them by basisically turning it into a gap-filling exercise. Gap-filling does not necessarily enhance critical thinking for academic phraseology; at NPP we're looking at a lot of potentially notable topics , especially among the popular BLP types, but in very poorly written articles that need a lot of attention before they are anywhere near ready for mainspace.

Thanks for the input @Kudpung.

@Pginer-WMF, yes, I do think subsets for bios could be considered but I would guard against creating too many types of article templates at least until some have been tried out for impact and run past UX and UI designers from the community's veterans first.

Makes sense. One of the aspects we want to learn from the initial experiment is which percentage of the article creation activity is covered by each outline. That can be useful to inform communities which are the key types of articles to cover the majority of the cases (or those more often leading to deletions) and at which point having more outlines may get diminishing returns. Ultimately, each community can determine which types of articles to support and to which extent.

The effort is to help the new, new page creators to create their articles rather than doing it for them by basisically turning it into a gap-filling exercise. Gap-filling does not necessarily enhance critical thinking for academic phraseology; at NPP we're looking at a lot of potentially notable topics , especially among the popular BLP types, but in very poorly written articles that need a lot of attention before they are anywhere near ready for mainspace.

The intention of the initial contents provided in the outlines is to serve as a specific contextually relevant example for the user to get started and get a practical sense of the information and style of writing that is expected. In some cases it may take more the form of "gap-filling" ("[Country name], officially [official name], is a country in [region]. "), in other cases, it can be more of a description of the type of contents expected ("[Description of the current governmental system and key institutions.]").
Excessive "templatization" can be a risk, and it is definitely worth observing whether exposure to these contents helps users get the sense of the underlying policies or they just stick too narrowly to filling in the information without further thinking.

@Pginer-WMF . The proposed list above of biography types is good. Each type has its own specific challenges of layout and content. Most of these types are written by fans and are often not of a suitable standard of language for including immediately in mainspace, or include text created by LLM. However, politicians present a different set of challenges: they are often written by professional public relations and political image consulting firms with focus on crafting a positive public persona. They generally bypass any help pages, and place articles directly into mainspace.

Hi,

I see you’ve already gone ahead anyway and created 48 page templates.
https://b24e11a4f1.catalyst.wmcloud.org/wiki/Category:Pages_using_article_guidance

  • That’s a few more than the 12 I recommended. A lot of work if they have to be coded up for 7 languages, and tested.

My theory was that if the new users were to be left to do some of the work themselves, it might have a greater didactic impact.
What’s the ETA on this project? And is it only being built for mobile platforms?
Even with 500 new articles a day, some of those templates would get used once a year at the most on en.Wiki.
Like the Newcomer Homepage, it’s also confronting the new user with information overload: Qual der Wahl (agonía de elección).

Regards,

Kudpung.

Hi,

I see you’ve already gone ahead anyway and created 48 page templates.
https://b24e11a4f1.catalyst.wmcloud.org/wiki/Category:Pages_using_article_guidance

  • That’s a few more than the 12 I recommended. A lot of work if they have to be coded up for 7 languages, and tested.

Outlines in the test instance were created (during late April) based on survey responses by Wikipedia editors. We created 40 outlines, and some editors added a few more. In any case, these represent a broad set of examples, and pilot wikis don't need to adapt them all.

As part of the information for pilot wikis to adapt outlines, we provided a list of 10 outlines to prioritize which was informed by input including yours.

What’s the ETA on this project? And is it only being built for mobile platforms?

The timeline for the project is to run the experiment during this quarter (April-June).

It is not built only for mobile platforms. The project works on the web across platforms. We decided to have mobile as the focus since it was an opportunity to make mobile contributions easier. However, the approach is not mobile-exclusive and the ideas applied are expected to be valid and helpful for those users on desktop. Along these lines, when we launch the experiment both mobile and desktop users will get the feature,.

Even with 500 new articles a day, some of those templates would get used once a year at the most on en.Wiki.
Like the Newcomer Homepage, it’s also confronting the new user with information overload: Qual der Wahl (agonía de elección).

Whether defining an outline is worth the effort for the community when it is used by a small percentage of articles is a totally valid question. I see this as a question for each community to answer.

For the current initiative, I think it is important to look at this from the experimental perspective. For the experiment I think it is useful to consider a broader set of outlines. Learning which are the most useful, to which degree, and for which wikis seems valuable information. Information that we cannot get if we preemptively discard outlines to only include the very essential.

I'd expect that users creating an article may be impacted the most by the availability or not of guidance for their article. For example, users trying to create an article about the recently discovered Bicharracosaurus will get guidance if there is an outline for "Fossil", for example. Someone trying to create that article may not be very impacted if there are several other outlines that do not apply to dinosaurs.