Wikimedia Meta-Wiki

WikiCite/2020 Virtual conference/The frontend of WikiCite

< WikiCite | 2020 Virtual conference
Open citations & linked bibliographic data | 26-28 October 2020 | #WikiCite
Part of Celebrating Wikidata's 8th Birthday | #WikidataBirthday

The frontend of WikiCite: How will the citation data actually be used in wiki projects? Templates, modules, and cross-project collaboration

edit

30min

Summary

edit

WikiCite is an initiative to develop open citations and linked bibliographic data to serve free knowledge, a series of conferences and workshops in support of that goal, and a community of people and ecosystem of projects which focuses on source metadata leveraging the Wikidata platform.

So, a conference is what we're doing right now. The technology to store open citations and linked bibliographic data will hopefully come in the future, and the community is being built around it. But once this technology is available, how will it actually be used by the wikis' editors of and displayed to the wikis' readers? This session will try to ask these relevant questions:

Video
 
Slides
  • How will the references be added by the wiki editors to the references repository? How will they be subsequently modified and, when necessary, deleted?
  • How will they be patroled and checked for hoaxes, misinformation, and other vandalism?
  • How will the references that are currently stored in millions of <ref> tags in the wiki projects move to this repository? Should they?
  • How will the references from the repository be inserted into pages on the different projects—Wikipedia, Wikisource, Wikibooks, and others?
  • What will be the wiki syntax for doing this? How will it be supported in Visual Editor, Content Translation, mobile apps, and other editing environments?
  • How will they be displayed to the readers?
  • How will they be reused across different wikis and languages? What should be similar and what should be different?
  • How can the editors of different wikis collaborate with each other most effectively on all of this?

We don't expect to answer all of these questions, but we certainly hope to start these very necessary conversations.

Bios

edit
  • Amir Aharoni (User:Amire80) is a linguist and software developer. He has edited Wikipedia, Wikisource and other projects in Hebrew English, Russian, and other languages since 2005. Since 2011 he's also a staff member of the Wikimedia Foundation (as User:Aaharoni-WMF), currently serving as Senior Strategist in the language team.
  • Andy Mabbett (User:Pigsonthewing) has been Wikipedia editor since 2003 and has been a Wikimedian in Residence at many GLEAM (Galleries, Libraries, Educators, Archives, Museums) institutions. He is currently in residence with ORCID and with Coventry University. He has given talks on Wikimedia subjects on five continents.
  • Add

Presentations

edit
edit

Speaker: User:Carn (engineer in ruwiki)

We want to use information from different types of sources (books, newspapers, magazines, articles or aggregators that provide them, sports databases, archives, materials from Wikisource, encyclopedias, law codes, tweets, etc), represent it in different standards or styles (GOST, CBE / CSE, Bluebook, MLA, MHRA, Chicago, BibTeX, LaTeX), link to our sources as accurately as possible many times to different source pages on one or different Wikipedia-pages. We want the content of the parameters to be easy for the user to enter and easy for the template to understand. It seems that we want everything and we want it now.

Making a small, specialized template is easier than making a big one or even a lua module. It is not customary to delete templates when they have different functionality if they are used. This leads to a whole zoo of different very similar templates ("forks"). Some templates have many parameters, others are limited to a few. When translating articles from English, German, French, Italian, or Spanish Wikipedia, users also copy templates or modify local templates in such a way that it is enough to change the name of the template and leave the "foreign" parameters. In one language section, a template may already be updated with a lua module, but its outdated copies will exist and work in other language sections of Wikipedia. By copying a template with parameters from one section to another, you can waste a lot of time not understanding why it doesn't work as intended.

Module:Citation/CS1 family of templates is widely used in ruwiki, but the community want to simultaneously receive technical updates from enwiki and maintain compliance of its output with common Russian-language source display standards. There are several specific templates developed within ruwiki. Here are some of them: {{Книга}} ("Kniga" = Book) and {{Статья}} ("Statja" = Article), but instead you may want to use more versatile template {{Публикация}} ("Publikatsiya" = Publication) or even {{Source-ref}} (powered by Module:Sources and information from Wikidata — {{Cite Q}} analogue). Different formats for presenting information affect the information itself. In enwiki first and last name are stored separately, in ruwiki they are molded together. Pages to which the citation goes indicated differently ("340-51" = "340-351"). For some editors, these small differences turn out to be very important.

The transfer of information to one place should be encouraged. Suppose that in some not very distant future we will solve the problem of incompatibility of different templates and we will establish a one-to-one correspondence between the parameters of the most popular templates in different language sections of Wikipedia with the values of properties in Wikidata, as well as we will clean up errors in the parameters, which at the moment do not critically affect the work of templates, but may lead to errors in the future. And we will take ref-records that are currently used on millions of Wikipedia pages in templates to populate a Wikidata with this information. If this information will be full, this will allow us to make sources citations using only one template and one shortcode - element id. But this is not enough. In addition to those sources that are already recorded in Wikipedia, we need a description of all books, articles, and newspapers from all libraries and databases. But how not to get confused in all this diversity? If the source is formatted without templates, sometimes we can automatically understand what it refers to in automatic mode if it uses codes ISBN, DOI, PMID, etc. We can do the same when detecting duplicate records and merge similar elements.

Most consensus practices can be taken from templates to MediaWiki extensions: WMDE Technical Wishes/Book referencing. But reaching consensus is a difficult process, we are not always able to agree on simple things quickly and without losses, for example, how many columns there should be in footnotes, how wide each column should be, should the link style be Harvard or not.

Editing awareness and trust experiments on iOS

edit

Speaker: Carolyn Li-Madeo (designer on the iOS team)

This year the iOS team is working towards the goals of closing the editing loop and breaking down the editing wall. As we begin to embark on this work, we would like to be able to define a few potential experiments for the article view in the iOS app, which will help readers gain an understanding of how Wikipedia works without losing trust in Wikipedia. The intended goal of this work is to empower readers to make corrections when they see mistakes or even potentially become successful editors.

One of our proposed experiments is focused on increasing reader awareness of high quality citations by injecting them into the article view. The goal of this experiment is to enable readers to learn more about how Wikipedia is built using reliable sources and to explore similar articles or resources.

Cite Q

edit

Speaker: Andy Mabbett, User: Pigsonthewing

Why citations in our beautiful, messy, global, and multi-lingual project should use centralised metadata (and why not doing so is stupid); what we need to do to get there, and how Cite Q is helping to deliver that.

Etherpad notes

edit

Introductory presentation

edit
  • Q: Has Citation Stylesheet Language (CSL - https://citationstyles.org/ ) been considered for use in Wikimedia Projects? if so, how?
    • Response from Amir after the session: I'm not familiar with it! Maybe other Wikicite people are familiar with it though. Thanks a lot for the pointer, this may become useful.
    • (More responses?)
  • Great point about 'why are references down at the bottom'! Several academic journals now put the references in a panel on the right which scrolls to the correct one when clicked, which would be ideal for WP. Examples: https://www.nature.com/articles/s41467-018-04669-9, https://onlinelibrary.wiley.com/doi/full/10.1002/prot.25679
    • Response from Amir after the session: Thanks! This is exactly the kind of thing I'm talking about. The iOS "Featured references" experiment, about which Carolyn speaks in her presentation has a similar spirit.
    • (More responses?)

Editing awareness and trust experiments on iOS

edit
edit
  • Comment about open library data: This discussion in Germany is quite old and positively settled in the sense as most the title data from Germans libaries is CC0, cf. e.g. http://openbiblio.net/
    • Response from Carn: For Russian libraries the set of bibliographic data is the subject of copyright. Although the data of each individual book is not protected.

Cite Q

edit
  • Questions:
    • - what does 'author-mask' look like?
      • can be seen on, for example, https://en.wikipedia.org/wiki/Peter_Bird_(IT_manager)#Bibliography with valueset to "0"; and in https://en.wikipedia.org/wiki/Emma_Louisa_Turner#Publications with the value "with" for a work written "with Gurney, Robert. "
    • - GoEThe: Will it possible to search for a particular reference work to get a particular QID, from the Wikipedia you want to use the Cite Q template?
      • Yes, via Citoid (or a Citoid-like feature)
    • - thoughts on a bot to move to Cite Q, for most used references? or semi-automated tool to incorporate custom page number citation.
      • Good idea, but likely to be subject of community pushback (especially on en.Wikipedia at the current time)
    • Q12345 is not very informative in wikitext - you wouldn't know in a minute which Q654321 is for which book. Can we use some human-friendly IDs?
      • No, but software should dipslay the title etc. in a pop-up, and can do so in the user's preferred language
Retrieved from "https://meta.wikimedia.org/w/index.php?title=WikiCite/2020_Virtual_conference/The_frontend_of_WikiCite&oldid=24909587"