Page MenuHomePhabricator

Autosuggest-Gadget does not find existing items in some cases
Open, LowPublicBUG REPORT

Description

Hello,

the tool was not able to find the existing object

https://www.wikidata.org/wiki/Q112519181

for article

https://de.wikipedia.org/wiki/Harald_Lange_(Sportwissenschaftler)

since it did not have a german label, but only an english label. In my opinion, this object should be found as well by the autosuggest gadget.


Another example:

AutoSuggest did not find the existing object

https://www.wikidata.org/wiki/Q95260054

for article:

https://de.wikipedia.org/wiki/Joachim_Fritsche_(Gebrauchsgrafiker)

It was possible to find the existing object by clicking on the Wikidata icon in the VIAF entry.


Thanks!

Event Timeline

Another example:

AutoSuggest did not find the existing object

https://www.wikidata.org/wiki/Q94776988

for article:

https://de.wikipedia.org/wiki/Erika_Klein_(Grafikerin)

It was possible to find the existing object by clicking on the Wikidata icon in the VIAF entry, though.

Another example:

AutoSuggest did not find the existing object

https://www.wikidata.org/wiki/Q109813191

for article:

https://de.wikipedia.org/wiki/Wilhelm_Zimmermann_(Mediziner)

It was possible to find the existing object by clicking on the Wikidata icon in the VIAF entry, though.

Another example:

AutoSuggest did not find the existing object

https://www.wikidata.org/wiki/Q94624318

for article:

https://de.wikipedia.org/wiki/Thomas_F._Naegele

The existing object could be found by clicking on the Wikidata icon in the VIAF entry.

AutoSuggest did not find the existing object

https://www.wikidata.org/wiki/Q95229883

for article

https://de.wikipedia.org/wiki/Karel_Petr_Adam

It was possible to find the existing object by clicking on the Wikidata icon in the VIAF entry, though.

AutoSuggest did not find the existing object

https://www.wikidata.org/wiki/Q12157666

for article

https://de.wikipedia.org/wiki/Bohdan_Stelmach

It was possible to find the existing object by clicking on the Wikidata icon in the VIAF entry, though.

AutoSuggest did not find the existing object

https://www.wikidata.org/wiki/Q99383843

for article

https://de.wikipedia.org/wiki/Alexandra_Martini_(Designerin)

It was possible to find the existing object by clicking on the Wikidata icon in the VIAF entry, though.

AutoSuggest did not find the existing object

https://www.wikidata.org/wiki/Q109806972

for article

https://de.wikipedia.org/wiki/Hans_Engels_(Maler)

It was possible to find the existing object by clicking on the Wikidata icon in the VIAF entry, though.

AutoSuggest did not find the existing object

https://www.wikidata.org/wiki/Q94943444

for article

https://de.wikipedia.org/wiki/Gustav_Beyer_(Pfarrer)

It was possible to find the existing object by clicking on the Wikidata icon in the VIAF entry, though.

AutoSuggest did not find the existing object

https://www.wikidata.org/wiki/Q55847847

for article

https://de.wikipedia.org/wiki/Karl_Wilhelm_Lin%C3%9F

It was possible to find the existing object by clicking on the Wikidata icon in the VIAF entry, though.

AutoSuggest did not find the existing object

https://www.wikidata.org/wiki/Q112470285

for article

https://de.wikipedia.org/wiki/Hermann_K%C3%BCster_(Mediziner)

It was possible to find the existing object by clicking on the Wikidata icon in the VIAF entry, though.

AutoSuggest did not find the existing object

https://www.wikidata.org/wiki/Q94891258

for article

https://de.wikipedia.org/wiki/Max_Pauer_(Bibliothekar)

It was possible to find the existing object by clicking on the Wikidata icon in the VIAF entry, though.

Regarding the correct links via VIAF entries: it sounds like the process on dewikipedia is for people to have researched the VIAF ID and added it to the wikitext in e.g. {{Normdaten|TYP=p|GND=117037133|VIAF=62316021}}. So the AutosuggestSitelink gadget would have to inspect the wiki page (either the wikitext or the HTML, although I note that the Normdaten template doesn't output any particularly useful HTML to identify the external IDs), and then query VIAF to see if a Wikidata item has been linked there. That feels like it might be an overly-specific workflow to be worthwhile: wouldn't it be a more general idea that anyone researching to find a VIAF ID be encouraged to instead do that work on Wikidata and so just add the sitelink?

One pattern I see in the page titles linked above is that a match might've been found if the parentheses were removed, so maybe that could be something that the gadget could do: search for the full page title, as well as a truncated version?

For other things such as Karl Linß on Wikidata not being matched to Karl Wilhelm Linß on dewikipedia, perhaps a similar approach could be taken: when there are multiple words, split them apart and search for all permutations.

I guess what I'm getting at is that it'd be better to make more general rules for the gadget to follow, rather than wiki-specific ones. After those rules have been implemented, then we could look at specific ones.

AutoSuggest did not find the existing object

https://www.wikidata.org/wiki/Q31190295

for article

https://de.wikipedia.org/wiki/Paul_Volzius

It was possible to find the existing object by clicking on the Wikidata icon in the VIAF entry, though.

The Autosuggest-Gadget stopped to work, since the full text search is not working since yesteray, for example

https://www.wikidata.org/w/index.php?go=Seite&search=Wilhelm+Fries&search=Wilhelm+Fries&title=Special%3ASearch&ns0=1&ns120=1

gives the error message

"Bei der Suche ist ein Fehler aufgetreten: Deine Suche konnte aufgrund eines vorübergehenden Problems nicht abgeschlossen werden. Bitte später erneut versuchen."

However, it is possible to find the existing object by clicking on the Wikidata icon in the VIAF entry, though, for example:

https://de.wikipedia.org/wiki/Wilhelm_Fries_(Maler)
https://www.wikidata.org/wiki/Q61775881

Also see
https://phabricator.wikimedia.org/T339814

The Autosuggest-Gadget stopped to work,

That's T339810 and unrelated to the Autosuggest-Gadget

AutoSuggest did not find the existing object

https://www.wikidata.org/wiki/Q116286696

for article

https://de.wikipedia.org/wiki/Andreas_Helfer

The wikidata object only had set the english label, but not the (identical) german label.

It was possible to find the existing object by clicking on the Wikidata icon in the VIAF entry, though.

Comment:

As for biographical articles a click in the VIAF entry helps to find the related wikidata object, for geographical objects the function "Special:Nearby"

https://www.wikidata.org/wiki/Special:Nearby#/coord/48.650833,16.679444

is very helpful to find the wikidata object, which might have various names in different languages.

For movies the IMDb ID can be used for the various movie titles in different languages, for chemical compound the CAS id can be used, for flight incidents the ASN id, for monuments the (country specific) monument ID, and so on.

Also see:
https://de.wikipedia.org/wiki/Benutzer:M2k~dewiki/FAQ#Wie_finde_ich_ein_bestehendes_Wikidata-Objekt_zu_einem_Artikel?

Hello,

what would be necessary to activate this tool for all users of a language version (e.g. the german language wikipedia) per default?

Is there any project plan or time schedule for testing and deploying?

Thanks a lot!

https://meta.wikimedia.org/wiki/Talk:AutosuggestSitelink

Regarding the correct links via VIAF entries: it sounds like the process on dewikipedia is for people to have researched the VIAF ID and added it to the wikitext in e.g. {{Normdaten|TYP=p|GND=117037133|VIAF=62316021}}. So the AutosuggestSitelink gadget would have to inspect the wiki page (either the wikitext or the HTML, although I note that the Normdaten template doesn't output any particularly useful HTML to identify the external IDs), and then query VIAF to see if a Wikidata item has been linked there. That feels like it might be an overly-specific workflow to be worthwhile: wouldn't it be a more general idea that anyone researching to find a VIAF ID be encouraged to instead do that work on Wikidata and so just add the sitelink?

The german language wikipedia is maintaining IDs for authority control (VIAF, GND, LCCN) locally, afterwards they can be transferred to wikidata using HarvestTemplates, so they can be used by all other language versions and projects (e.g. commons). Also see:

Thanks for the links, although I'm still not sure I'm understanding: what particular changes do you think AutosuggestSitelink needs in order to work better for German Wikipedia?

From what I read, it sounds like dewiki is supportive of using data from Wikidata in the Nordaten template, and is doing so for quite a lot of them. Are the manually-entered ones generally the older ones, done before the Wikidata integration was added? As you mention above many times, there is already a Wikidata item for lots of pages where someone has added the VIAF ID to both the Wikdiata item and the text of the page. However, there's no easy way for the gadget to know that these match.

Thanks for your help, and I'm sorry if I'm not understanding properly!

JWheeler-WMF subscribed.

@HMonroy do we know how we might fix this issue?