Wikidata:WikidataCon 2017/Notes/Wikidata-powered infoboxes on the Basque, Catalan and English Wikipedias

Title: Wikidata-powered infoboxes on the Basque, Catalan and English Wikipedias.

This session gathers three similar submissions with different experiences of how to deploy Wikidata-powered infoboxes.

Note-taker(s): Andrew Lih (User:Fuzheado)

Speaker(s) edit

Name or username: User:Mike Peel + User:amadalvarez User:Paucabot + User:Theklan

Session slides on Commons: https://www.wikidata.org/wiki/File:Wikidatacon_2017._Wikidata-powered_infoboxes.pdf

Intersting tools/pages:

Transclusion checker tool, showing French Wikipedia use of Infobox Monuments (19,876 uses)

https://tools.wmflabs.org/templatecount/index.php?lang=fr&namespace=10&name=Infobox+Monument#bottom

Catalan infobox examples: https://ca.wikipedia.org/wiki/Viquiprojecte:Adaptaci%C3%B3_de_plantilles_a_Wikidata/infotaules/Mostres'

Abstract edit

This panel discussion brings together the experiences from including Wikidata information into the infoboxes from three Wikipedias. We will each give a presentation of 5-10 minutes, followed by an audience Q&A. It was originally proposed as three separate sessions, the original abstracts for which are below.

Basque Wikipedia

Wikidata is a powerful tool for small language communities. We can invoke Wikidata and have up-to-date information on a huge variety of topics, reducing the time we waste on template maintenance and centering ourselves on article edition. It's a while since we started to have fully automatic templates in the Basque Wikipedia and the article creation workflow has been dramatically increased, while the templates are now more coherent, better organized and with standard labels. As a small community, this is like a dream that can only go better, as more data will always be available.

Catalan Wikipedia

When we started 2 years ago implementing wikidata to existing infoboxes, we realized that it was huge job because we had hundreds of hyperspecialized infoboxes (regions, professions, church, castle, building, ...). Then, we decided to build core infoboxes that gathered the best features and parameters from all the infoboxes with a similar topic. These new infoboxes work with intense use of wikidata in addition to the same manual parameters they had. De facto, the publishers have stopped entering manual parameters, it is easy to use, and the guarantee of maintenance of volatile contents is higher than before.

Today, 10 "core infoboxes" serve 55% of all cawiki articles and have left behind as obsolete almost 300 infoboxes. All of them include dynamic maps based on OSM (when applied), multivalued properties integrating multiple qualifiers in the edition of the property. Additionally, there are 30-40 highly specialized infoboxes -attending 5-10% of cawiki articles- that have not been able to concentrate, but have been powered with the same wikidata functionalities.

This project has required the homogenization of the names of the parameters that represent the same concept, creating a kind of data dictionary of manual parameters and then, rename the old parameter names in articles in order to run with the new infobox. Our Wikidata module has evolved and, at the moment, it is able to do small editing treatments for the combination property-qualifiers; Manage the gender to return correctly the description of occupations or positions when women; Recover bottom-up hierarchical structures such as taxon or higher territorial units, etc.

The empiric experience allows audience to see (and copy the specific solution) how to handle hundreds of properties.

English Wikipedia

Data from Wikidata is now being used live in a number of infoboxes on the English Wikipedia to display pictures, coordinates, links, numbers with automatically-converted units, location maps, even references. Over 600 700 800 900 1,000 infoboxes are completely Wikidata-driven, and many more are partially Wikidata-driven. A good live example of this is the en:South Pole Telescope, where everything shown in the infobox is fetched from Wikidata (see the picture on the right).

We aren't doing this perfectly yet - there are limitations in the data storage and the programming logic that need to be improved. Community consensus about how to appropriately include Wikidata information is also very important here.

This session will cover the technical process of including Wikidata information on the English Wikipedia (through en:Module:Wikidata and en:Module:WikidataIB), the technical limitations that have been encountered whilst doing so, and the social issues that are also present in this transition. It will also cover the different approaches to Wikidata infoboxes across the different language Wikipedias.

Collaborative notes of the session edit

Mapping is done automatically -

Single points

point to point or path - Boudewijnkanaal, Pedraforca

streets - use "intersects with" to draw a path through number of points - Paseo de Gracia

Example page: https://ca.wikipedia.org/wiki/Viquiprojecte:Adaptaci%C3%B3_de_plantilles_a_Wikidata/infotaules/Mostres

58% of articles in CA use wikidata infoboxes!

More wikidata-powered infoboxes → more info to Wikidata

Basque:

   Small communities cannot maintain templates up to date
   No learning curve - insert and it's done. Easier for newbies, non tech users, librarians, people who want to write
   Translate on the fly - insertion of infobox discovers what has and has not been translated
   20,000 biographies
   USE IT! COPY IT! IMPROVE IT! https://labur.eus/ohWTFs
   

Mike Peel - EN.WP

"So on English Wikipedia, we have something very different..."

<Rounds of laughter>

https://en.wikipedia.org/wiki/Template_talk:Infobox_World_Heritage_Site#RfC:_revert_back_to_non-Wikidata_version.3F

  1. invoke:wd to display reference URL

English Wikipedia expectation of undoing vandalism very quickly may be holding it back from adopting Wikidata, where this is not handled immediately.

Someone changed France to Germany - took 10 hours to fix, whereas en.wp wants it faster

en.wp wants it to be perfect before it is implemented, or at least better than the best infoboxes