Wikidata:WikiProject LD4 Wikidata Affinity Group/Affinity Group Calls/Meeting Notes/2022-08-23

From Wikidata
Jump to navigation Jump to search

Call details[edit]

Presentation material[edit]

No presentation materials were provided

Notes[edit]

  • Diego de la Hera and Scann on Web2Cit, a tool to collaboratively improve automatic citations in Wikipedia
  • Wikipedia has automated citation tool, Citoid, if you have a reference to generate a citation.
  • Sometimes it doesn’t work - date may only be the retrieval date, names and title could be wrong, etc.
  • Webpage metadata needs to be structured in the same way for the metadata to be extracted correctly or each webpage needs its own translator.
  • Study: Citoid is only returning hits for certain components: but only 31.9% had all four components correct.
  • Preliminary results: only 48% correct responses.
  • How to fix:
    • Manually in Wikipedia, sometimes relatively easy, sometimes cumbersome
    • Fix the metadata on the webpage: best way to go, although might take some discussion and time.
    • Fix the translator (Zotero): but would need to know programming language (javascript): also involves some bureaucracy to wait till Zotero team approves & Wikimedia team incorporates.
  • Web2Cit: Financed by Wikimedia Foundation through grant. Diego is project manager along with a team of others.
  • What does Web2Cit do: Uses combination of Citoid and other resources (potentially including Wikidata):
    • Define and maintain translation procedures & tests, simply & collaboratively
  • Web2Cit configuration (see the recording for a demo)
    • Will configure on a per domain (website)
    • Q: Do you use specific citation style?
      • A: Operates like Citoid: depends on Wikipedia. Web2Cit provides the fields, not the citation style/structure.
    • Will also have translation tests and templates
    • User script can be installed from meta.wiki, https://meta.wikimedia.org/wiki/Web2Cit, bottom of the page “Integrating Web2Cit in Wikipedia”, links to instructions on how to install the script to your personal Wikipedia account.
    • Now “Add a citation” box in Wikipedia will provide 2 citations, one from Citoid and one from Web2Cit.
    • Translation summary of the URL we are working with is available here: https://web2cit.toolforge.org/+https://edition.cnn.com/2022/08/22/world/jupiter-images-webb-telescope-nasa-scn/index.html  
    • In Web2Cit, we want to define a test for a specific path (webpage) by defining test fields and expected values under the “Expected Output” of the Translation Summary page. Once saved, a JSON file is saved to Wikipedia. Web2Cit will give a score of Citoid result vs. Expected result.
    • Question: What happens when the website disappears or the URL changes–is there anyway to add that change?
      • A: Best solution might be to use Internet Archive, the Wayback Machine.
    • Once a test is set up, we can edit the Translator. We can set Translation Procedures to provide the correct metadata. It may be that we want to select “Fixed Selection” and a specific configuration (ex. authorLast) or we can use the Citoid selection if it was correct. Once saved, the JSON file is updated in Wikipedia and we can see an updated Translation Summary page in Web2Cit.
    • Power of Web2Cit is that it created a Translation Template - the template can be used for any other pages from that domain (website) and it would hopefully be correct.
    • Multiple translation templates: Can define more than one template for website and multiple translation subgroups (can read documentation for further reference)
    • There is a proposal for Citoid to support automatic references in Wikidata. Citoid integration (T199197).  Is working in beta form and not sure where this project is.
    • CiteTool - could be tweaked to use Web2Cit
    • Automated item creation: Wikicite community: wouldn’t it be great if you had a DOI that would create an item on Wikidata.
      • Can also import through Zotero using QuickStatements.
    • Can also use Zotero with the Web2Cit by clicking the Save to Zotero web plug-in tool while on the Translation Summary page on Web2Cit.

Questions[edit]

  • Q: Does the template work across users?
    • A: Yes, Web2Cit is a collaborative tool and everyone who has authorization can share and contribute.
  • Q: Asks about website that was free but now behind a paywall.
    • A: Wikipedia has a bot to test these links.
  • Q: Where can I find CiteTool?
  • Web2Cit Userguide: https://meta.wikimedia.org/wiki/Web2Cit/User_guide
    • Find videos or how Web2Cit works, the Web2Cit ecosystem, and How to use Web2Cit
  • Web2Cit research: https://meta.wikimedia.org/wiki/Web2Cit/Research
  • Q: Do you have an estimate of how many domains are represented in Wikipedia citations?
    • A: Research team worked on featured articles only. We only have estimated for these featured articles: 300,000 citations have been extracted from English, Spanish & French Wikipedias.
  • If you are interested in learning about this tool, can look at translation templates that have been defined for other tools. This website might give you an idea of which websites have configuration profiles setup: https://meta.wikimedia.org/wiki/Special:PrefixIndex/Web2Cit/data
    • Q: Why are some items in list as italics and grey?
      • A: Those are redirects, sometimes websites are aliases/equivalent, so it doesn’t make sense to define two separate configuration profiles.
Retrieved from "https://www.wikidata.org/w/index.php?title=Wikidata:WikiProject_LD4_Wikidata_Affinity_Group/Affinity_Group_Calls/Meeting_Notes/2022-08-23&oldid=2119442691"