User:PAC2/Signpost Opinion1
Are Wikipedia articles representative of Western or world knowledge?
Article display preview: | This is a draft of a potential Signpost article, and should not be interpreted as a finished piece. Its content is subject to review by the editorial team and ultimately by JPxG, the editor in chief. Please do not link to this draft as it is unfinished and the URL will change upon publication. If you would like to contribute and are familiar with the requirements of a Signpost article, feel free to be bold in making improvements!
|
YOUR ARTICLE'S DESCRIPTIVE TITLE HERE
Optional: write a lede — not necessarily a WP:LEAD. Interesting > encyclopedic.
Wikipedia aims at representing the sum of all knowledge. It is not so easy to define "the sum of all knowledge". We could expect the sum of all knowledge means knowledge from every region in the world (geographical distribution), from every era in History, from every culture, every ethnic group every gender group, etc.
Trying to measure diversity of knowledge in Wikipedia, we can look at diversity of contributors, number of Wikipedia articles, diversity of sources and references[1] or diversity in mentioned entities inside a given article.[2]
In this article, I look at the geographical distribution of people mentioned in an article (people mentioned with a blue link).
I apply my methodology to a selection of articles about general topics such as music, politics, culture or religion in a selection of Wikipedia versions and I discuss the results.
Methodology
Given a Wikipedia article, I select all internal links (blue links) and I call them "mentioned entities". This can be done through the endpoint "links" in the MediaWiki generator API. The magic is that this API can be integrated in a SPARQL query in the Wikidata Query Service. So I combine the call to the API with a Wikidata query. I select all mentioned entities with P31 equal to Q5 (humans) with a known birthplace (P19) and I collect the country of the birthplace with property P17.
SELECT DISTINCT ?item ?itemLabel ?country ?countryLabel ?birthplace
?birthplaceLabel
WHERE {
SERVICE wikibase:mwapi {
bd:serviceParam wikibase:endpoint "en.wikipedia.org";
wikibase:api "Generator";
mwapi:generator "links";
mwapi:titles "Music";.
?item wikibase:apiOutputItem mwapi:item.
}
FILTER BOUND (?item)
?item wdt:P31 wd:Q5 ; wdt:P19
?birthplace.
?birthplace wdt:P17 ?country .
SERVICE wikibase:label { bd:serviceParam wikibase:language "en,mul". }
}
I then collect a mapping between actual countries and continents. The mapping comes from Wikidata but is consistent with United nations M49 classification[3].
SELECT DISTINCT ?continent ?continentLabel ?country ?code WHERE {
VALUES ?continent {
wd:Q55643
wd:Q48
wd:Q15
wd:Q18
wd:Q49
wd:Q46
}
?continent (wdt:P527*) ?country.
?country
wdt:P2082 ?code.
SERVICE wikibase:label { bd:serviceParam wikibase:language "en,mul". }
}
I perform a left join of the two data frames using the Arquero JavaScript library[4].
Finally, I regroup Europe and North America as "Western World" and the four other continents as "Rest of the world". This is an opiniated and radical approach but it makes the numbers easier to read. Places of birth which cannot be associated with a current country are labeled "Unclassified".
I've developed a user interface using Observable notebook.[5] Users can choose two parameters: the Wikipedia project (ie "pt.wikipedia.org") and the title of the article. Parameters can be added in the url directly. For instance, you can look at article "Kennis" (ie knowledge) in Afrikaans: https://observablehq.com/@pac02/wwrw?wikipedia=af.wikipedia.org&article=Kennis.
All computations are performed in the appendix of the notebook. The code is open source licensed under the ISC.
Results
This approach makes sense for articles about general topics such as music, work, art, beauty, love, humanity, knowledge, education, school, religion, etc. Also it makes sense if the number of people mentioned in the article is high enough to compute percentages.
In this section, we focus in three notions in English, French, Spanish and Arab.
Music
Music comes from all over the world. I would expect an encyclopedic article to mention people from all the continents. Let's have a look at the numbers.
Linguistic version | Article | Rest of the world | Unclassified | Western world |
---|---|---|---|---|
English | Music | 6 (8.2%) | 2 (2.7%) | 65 (89.0%)[6] |
Spanish | es:Música | 0 (0%) | 3 (5.0%) | 57 (95.0%)[7] |
Discussion
- ^ For instance, Piotr Konieczny and Włodzimierz Lewoniewski look at the number and of articles related to the United States of America and the number of American sources in references. See their presentation at Wikimania 2024: https://prezi.com/view/C7snnAZFWqZz7vPD0kLu/
- ^ In a previous Signpost article, I look at the gender distribution of people (human entities) mentioned in Wikipedia articles: Measuring gender diversity in Wikipedia articles, The Signpost, may 2022
- ^ https://unstats.un.org/unsd/methodology/m49/overview/#
- ^ Arquero is JavaScript library developed by Jeffrey Heer: https://idl.uw.edu/arquero/api/
- ^ Observable is a platform created by Melody Meckfessel and Mike Bostock which proposes to write notebooks in JavaScript. It is widely used by the data visualization community
- ^ https://observablehq.com/@pac02/wwrw?wikipedia=en.wikipedia.org&article=Music
- ^ https://observablehq.com/@pac02/wwrw?wikipedia=es.wikipedia.org&article=M%C3%BAsica
Discuss this story