Page MenuHomePhabricator

VE citation tool : frequent incorrect 3rd ISBN with no obvious reason for Worldcat
Closed, ResolvedPublic

Description

I regularly find articles with completely incorrect isbn3 in templates Ouvrage inserted by VE on frwiki that don't seem related to the actual book. It's always isbn3, not isbn, isbn2 or other parameter of the template. Examples of fixing the article: Musée René Magritte, 7 arts (revue d'avant-garde), but there were many other.

I've struggled to reproduce it several times, and finally found a user that gave me enough information to reproduce it.

  • Edit Musée Réné Magritte on frwiki with VE
  • Click on the "Sourcer" button
  • Type "jacques roisin magritte biographie" and click on the "Générer" button
  • 2 sources are suggested, and the second one is the citation with the problem :
<ref>{{Ouvrage|nom1=Roisin, Jacques.|titre=Ceci n'est pas une biographie de Magritte : la première vie de l'homme au chapeau melon|éditeur=Alice Éditions|date=1998|isbn=2-930182-05-9|isbn2=978-2-930182-05-6|isbn3=2002420539|oclc=43991516|lire en ligne=http://worldcat.org/oclc/43991516|consulté le=2020-04-04}}</ref>

There are 3 ISBN in the citation proposed by VE, the third one being completely incorrect. And I don't know from where it comes : if you look at Worldcat for the OCLC of the citation, only the 2 correct ISBN are visible, there's no mention of the third incorrect one. Where did the citation tool find this third ISBN?

I checked through the API also with the same request jacques roisin magritte biographie, and see the same problem in the JSON result:

[
  {
    "itemType":"journalArticle",
    ....,
    "source":["Crossref"]
  },
  {
    "itemType":"book",
    "title":"Ceci n'est pas une biographie de Magritte : la première vie de l'homme au chapeau melon",
    "url":"http://worldcat.org/oclc/43991516",
    "ISBN":["2-930182-05-9","978-2-930182-05-6","2002420539"],
    "oclc":"43991516",
    "date":"1998",
    "publisher":"Alice Éditions",
    "author":[["","Roisin, Jacques."]],
    "accessDate":"2020-04-06",
    "source":["WorldCat"]
  }
]

In the result I see that the source is WorldCat, but when I check the WorldCat URL, I can only see the first 2 ISBN, and there's nowhere a 3rd ISBN... Where does Citoid find this 3rd ISBN, which is completely incorrect?

Event Timeline

Other example on Cheval en Suède, with another incorrect 3rd ISBN "2014010379", and nothing in the WorldCat page shows that ISBN.

NicoV renamed this task from VE citation tool : frequent incorrect isbn3 with no obvious source to have this 3rd ISBN to VE citation tool : frequent incorrect 3rd ISBN with no obvious reason for Worldcat.Apr 6 2020, 10:08 AM

This sounds like a bug, I'll look into it.

Mvolz triaged this task as Medium priority.
JTannerWMF subscribed.

Looks like @Mvolz is working on this so I am moving this to Triaged

Do you have any other examples? When I check this now I'm only seeing the two (correct) ISBNs. Maybe it was a temporary bug from Worldcat upstream?

@Mvolz I've seen this bug for a long time (several months) and still seeing it on a regular basis, and always the same problem 3rd ISBN completely wrong and with Worldcat.

I've just fixed another one, the citation was added 2 days ago.

@Mvolz

With the previous example, I tried using the citation tool : when I use "Istorijsko poreklo Srba" as the search critera, I got 2 suggestions, the second one having the 3rd incorrect ISBN.

And if try the API with the same criteria, I see the same problem :

[{"itemType":"journalArticle","issue":"5","DOI":"10.4312/dela.5.1-11","pages":"1–11","title":"Poreklo prebivalstva slovenskih mest","publicationTitle":"Dela","date":"1987-12-01","url":"http://dx.doi.org/10.4312/dela.5.1-11","ISSN":["1854-1089","0354-0596"],"accessDate":"2020-06-04","author":[["Mirko","Pak"]],"source":["Crossref"]},{"itemType":"book","title":"Istorijsko poreklo Srba","url":"http://worldcat.org/oclc/792988023","ISBN":["978-86-911149-2-3","86-911149-2-4","2012467583"],"oclc":"792988023","author":[["","Milanović, Miodrag."]],"accessDate":"2020-06-04","source":["WorldCat"]}]

Change 602681 had a related patch set uploaded (by Mvolz; owner: Mvolz):
[mediawiki/services/citoid@master] Make ISBN validation stricter for open search

https://gerrit.wikimedia.org/r/602681

Change 602681 merged by jenkins-bot:
[mediawiki/services/citoid@master] Make ISBN validation stricter for open search

https://gerrit.wikimedia.org/r/602681