An combined R package for reading, writing and handling Wikidata semantic data (via APIs).
Authors: Thomas Shafee (aut., maint.), Os Keys (aut., cre.)
License: MIT
Status: Stable
WikidataR includes functions to: - read from wikidata (single items, properties, or properties) - query wikidata (retrieving all items that match a set of criterial via Wikidata SPARQL query service) - write to Wikidata (adding new items or statements via QuickStatements) - Handle and manipulate Wikidata objects (as lists and tibbles) For details on how to best use it, see the examples below.
To download WikidataR from CRAN:
install.packages("WikidataR","WikidataQueryServiceR")
To get the current development version from github:
install.packages("devtools")
devtools::install_github("r-lib/httr")
For cases where you don’t already know the QID of an item or the PID of a property, you can search wikidata by name. Note that some search terms will return multiple possible items. You can also specify a language (defaults to Engligh).
Which returns the lists:
acetaminophen (Q57055) - common drug for pain and fever
Paracetamol (Q36716177) - scientific article published on July 1980
Paracetamol (Q54982056) - musical group
...
and
medical condition treated (P2175) - disease that this pharmaceutical drug, procedure, or therapy is used to treat
Elements within those lists include basic information from wikidata (ID, description, labels). The QID or PID can then be used to get the full data for the item (see below).
Wikidata is an excellent thesaurus for different identifiers. For example it’s possible to convert from any identifier to wikidata QIDs or between different identifiers
qid_from_identifier('ISBN-13','978-0-262-53817-6')
identifier_from_identifier('ORCID iD','IMDb ID',c('0000-0002-7865-7235','0000-0003-1079-5604'))
Which returns the lists:
978-0-262-53817-6 Q102035721 Wikipedia @ 20: Stories of an Incomplete Revolution
and
# A tibble: 2 x 2
value return
<chr> <fct>
1 0000-0002-7865-7235 nm2118834
2 0000-0003-1079-5604 nm1821217
In this example, we search for three articles using their DOIs (P356), find their QIDs, download their full wikidata entries, and then extract the “main topics” (note PID didn’t have to be used).
article.qid <- qid_from_DOI(c('10.15347/WJM/2017.007','10.15347/WJM/2019.001','10.15347/WJM/2019.007'))
article.q <- get_item(article.qid)
article.topics.p <- extract_claims(article.q, "main topic")
get_names_from_properties(article.topics.p)
Which returns a tibble for each of the journal articles, listing the main topics of each and their QIDs.
$`10.15347/WJM/2017.007`
# A tibble: 1 x 2
QID value
<chr> <chr>
1 P921.Q164778 rotavirus
$`10.15347/WJM/2019.001`
# A tibble: 2 x 2
QID value
<chr> <chr>
1 P921.Q15989108 Western African Ebola virus epidemic
2 P921.Q10538943 Ebola virus
$`10.15347/WJM/2019.007`
# A tibble: 2 x 2
QID value
<chr> <chr>
1 P921.Q1820650 readability
2 P921.Q16235120 health information on Wikipedia
In this example, we search Wikidata for any items that are an “instance of” (P31) “film” (Q11424) that has the label “The Cabin in the Woods” (Q45394), and ask for the item’s genres (P136).
query_wikidata('SELECT DISTINCT
?genre ?genreLabel
WHERE {
?film wdt:P31 wd:Q11424.
?film rdfs:label "The Cabin in the Woods"@en.
?film wdt:P136 ?genre.
SERVICE wikibase:label { bd:serviceParam wikibase:language "en". }
}')
Which returns a tibble:
# A tibble: 6 x 2
genre genreLabel
<chr> <chr>
1 http://www.wikidata.org/entity/Q3072049 zombie film
2 http://www.wikidata.org/entity/Q471839 science fiction film
3 http://www.wikidata.org/entity/Q859369 comedy-drama
4 http://www.wikidata.org/entity/Q1342372 monster film
5 http://www.wikidata.org/entity/Q853630 slasher film
6 http://www.wikidata.org/entity/Q224700 comedy horror
For more example SPARQL queries, see this page on Wikidata.
query_wikidata()
can accept multiple queries, returning a (potentially named) list of data frames. If the vector of SPARQL queries is named, the results will inherit those names.
In this example we’ll write directly to wikidata via the QuickStatements format.
write_wikidata(items = c("Q4115189","Q13406268"),
properties = "author",
values = c("Q762","Q41406"),
format = "api",
api.username = "myusername", # Enter your Wikimedia username here
api.token = "" #REDACTED# Find yours from https://tools.wmflabs.org/quickstatements/#/user
)
Results in the statements being directly added to wikidata under your username via the API.
> The Mona Lisa (Q12418) has the Creator (P170) of Leonardo da Vinci (Q762)
> The Scream (Q471379) has the Creator (P170) of Edvard Munch (Q41406)
Alternatively, you can print via format=tibble
and paste into the QuickStatements website.
The example below finds all articles in a journal, works out the URL for their peer reviews, and writes those URLs into those articles’ wikidata items.
sparql_query <- 'SELECT ?Article ?ArticleLabel ?JLabel ?T ?peer_review_URL WHERE {
SERVICE wikibase:label { bd:serviceParam wikibase:language "[AUTO_LANGUAGE],en". }
?Article wdt:P1433 wd:Q24657325.
OPTIONAL { ?Article wdt:P1433 ?J. }
OPTIONAL { ?Article wdt:P1476 ?T. }
OPTIONAL { ?Article wdt:P7347 ?peer_review_URL. }}
LIMIT 10000'
articles.qr <- as_tibble(query_wikidata(sparql_query))
articles.qr <- articles.qr[articles.qr$peer_review_URL=="",] #omit those with review URLs listed
review.URLs <- paste0('https://en.wikiversity.org/wiki/Talk:',
articles.qr$JLabel,
"/",
articles.qr$T
)
review.URLs <- gsub(" ","_",review.URLs)
write_wikidata(items = sapply(sapply(articles.qr$Article,pattern = "/",stringr::str_split),tail,1),
properties = "Peer review URL",
values = review.URLs,
format = "tibble",
)
write_wikidata(items = sapply(sapply(articles.qr$Article,pattern = "/",stringr::str_split),tail,1),
properties = "Peer review URL",
values = review.URLs,
format = "api",
api.username = "myusername",
api.token = , #REDACTED# Find yours from https://tools.wmflabs.org/quickstatements/#/user
)
This package combines and builds on the utilities of Os Keyes’ WikidataR, Christian Graul’s rwikidata, Mikhail Popov’s WikidataQueryServiceR, and Serena Signorelli’s QueryWikidataR packages. It also uses the Magnus Manske’s QuickStatements tool.