Help

SPARQL

Revision as of 11:29, 9 December 2021 by Yug (talk | contribs) (→‎Merging data)

Base

Code snippets

Fetch data using SPARQL

LinguaLibre data can be fetched using various coding languages such as Python, Javascript, R and others, returning JSON or other formats.

  • For code snippet in your language : open query.wikidata.org (WikiData Query Service, aka WDQS), run your SPARQL query, click "Code" : a pop up window appears with various implementations.
  • For downloading data, click "Download".

Javascript:
At least 3 methods exists (code snippet), example:

Query Result's basic unit
SPARQL:
SELECT ?item WHERE { ?item prop:P2 entity:Q5 } LIMIT 10
{  },
{
  "item": {
    "type": "uri",
    "value": "https://lingualibre.org/entity/Q12"
  },
  "itemLabel": {
    "xml:lang": "en",
    "type": "literal",
    "value": "beginner"
  }
},
{  }
Javascript:
var endpoint = 'https://lingualibre.org/sparql';
var sparql = 'SELECT ?item WHERE { ?item prop:P2 entity:Q5 } LIMIT 10';
$.getJSON(endpoint,
	{ query: sparql, format: 'json' },
	function(data){ console.log('JQuery: ',data)}
);

Merging data

Advanced SPARQL queries with COUNT() and others are often slow (>3secs, sometime >100secs). You are encouraged to do multiple smaller SPARQL queries to then merge their responded data. By example, the complementary Javascript snippet below would help web developers to do so.

// Data from 3 sparql queries.
// Important: One key must be similar in all datasets, here: 'qid'
const langs = [{ qid: 'Q209', label: 'Breton', iso:'bre' }, { qid: 'Q21', label: 'French', iso: 'fra' }],
    speakersFemales = [{ qid: 'Q209', genderF: 3, recordsF: 60 }, { qid: 'Q21', genderF: 21, recordsF:15046 }],
    speakersMales = [{ qid: 'Q209', genderM: 7, recordsM: 112 }, { qid: 'Q21', genderM: 85, recordsM:82964 }];
// Toolbox for merging data by same id
var merge2ArraysBySameId = function(arr1,arr2,id1){
	return arr1.map( item1 => { 
  	var identical = arr2.find(obj => obj[id1] === item1[id1]); 
  	return Object.assign(identical, item1) 
  } );
}
// Mergings
var step1 = merge2ArraysBySameId(langs,speakersFemales,'qid');
var step2 = merge2ArraysBySameId(step1,speakersMales,'qid');
alert(JSON.stringify(step2))

Lingualibre's ground

✅ Is Language (Q4) → List existing languages

SELECT ?lang ?iso ?langLabel WHERE {
  ?lang prop:P2 entity:Q4 .
  ?lang prop:P13 ?iso .
  SERVICE wikibase:label {
    bd:serviceParam wikibase:language "[AUTO_LANGUAGE],en" .
  } 
}
... Loading ...

✅🇶 Is Speaker (Q3) → List existing speakers

SELECT ?speaker ?speakerLabel
WHERE {
  ?speaker prop:P2 entity:Q3 .  # Condition 1, P2 'instance of' is Q3 'speaker'.
  SERVICE wikibase:label {
    bd:serviceParam wikibase:language "[AUTO_LANGUAGE],en" .
  } 
}
... Loading ...

✅ Is Language level (Q5) → List existing levels

SELECT ?item ?itemLabel
WHERE {
  ?item prop:P2 entity:Q5    # Condition 1, P2 'instance of' is Q5 'language level'.
  SERVICE wikibase:label {
    bd:serviceParam wikibase:language "[AUTO_LANGUAGE],en" .
  } 
}
... Loading ...

✅ Is Sex or Gender (Q7) → List existing sexes or genders

SELECT ?item ?itemLabel
WHERE {
  ?item prop:P2 entity:Q7    # Condition 1, P2 'instance of' is Q7 'sex or gender'.
  SERVICE wikibase:label {
    bd:serviceParam wikibase:language "[AUTO_LANGUAGE],en" .
  } 
}
... Loading ...

Speaker

✅ Speaker name(s) → Speaker Qid(s)

SELECT ?speakerName ?speakerId
WHERE {
  VALUES ?speakerName { "Yug" "VIGNERON" } # One or multiple values
  BIND ( STRLANG(?speakerName, "en") AS ?speakerLabel )
  # P2: instance of; Q3: speaker.
  ?speakerId prop:P2 entity:Q3 ; rdfs:label ?speakerLabel .
}
... Loading ...

✅🇶 Speaker Qid (Q42) → Speaker data, all

# Get Q42 (User:0x010C)'s data
SELECT ?predicate ?object ?objectLabel
WHERE {
  entity:Q42 ?predicate ?object .
  SERVICE wikibase:label {
    bd:serviceParam wikibase:language "[AUTO_LANGUAGE],en" .
  } 
}
... Loading ...

✅🇶 Speaker Qid (Q42) → Speaker languages (P4)

SELECT ?languages ?languagesLabel
WHERE {
  entity:Q42 prop:P4 ?languages .
  SERVICE wikibase:label {
    bd:serviceParam wikibase:language "[AUTO_LANGUAGE],en" .
  } 
}
... Loading ...

✅ Speaker Qid + language Qid → List associated audios

SELECT ?audio ?audioLabel
WHERE {
  ?audio prop:P5 entity:Q42 .   # Condition 1, P5 Speaker is Q42 User:0x010C
  ?audio prop:P4 entity:Q21 .   # Condition 2, P4 language is Q21 French
  # Labels
  SERVICE wikibase:label {
    bd:serviceParam wikibase:language "[AUTO_LANGUAGE],en" .
  } 
}
... Loading ...

Languages

✅ Language LL Qid (Q21) → Count items

SELECT ?language (COUNT(?audio) AS ?nbAudio) WHERE {
  VALUES ?language { entity:Q21 }
  ?audio prop:P4 ?language .
}
GROUP BY ?language
... Loading ...

✅ Language LL Qid (Q21) → Count records

SELECT ?language (COUNT(?audio) AS ?audio) WHERE {
  VALUES ?language { entity:Q21 }
  ?audio prop:P2 entity:Q2 .  # P2 'instance of' is Q2 'record'
  ?audio prop:P4 ?language .  # P4 'language' is Q21 'French'
}
GROUP BY ?language
... Loading ...

?✅ Language LL Qid (Q21) → Count unique words

SELECT ?language (COUNT(?audio) AS ?Audios) (COUNT(DISTINCT(?itemLabel)) AS ?Words)
(ROUND(10000*?Words/?Audios)/100 AS ?Percent)
WHERE {
  VALUES ?language { entity:Q21 }
  ?audio prop:P2 entity:Q2 .  # P2 'instance of' is Q2 'record'
  ?audio prop:P4 ?language .  # P4 'language' is Q21 'French'
  ?audio rdfs:label ?itemLabel. # Assign value: label to ?itemLabel
}
GROUP BY ?language
... Loading ...

✅ Language LL Qid (Q21) → Count speakers

SELECT ?language (COUNT(?audio) AS ?audio) WHERE {
  VALUES ?language { entity:Q21 }
  ?audio prop:P2 entity:Q3 .  # P2 'instance of' is Q3 'speaker'
  ?audio prop:P4 ?language .  # P4 'language' is Q21 'French'
}
GROUP BY ?language
... Loading ...

✅ Language LL Qid (Q209) → List speakers

SELECT ?language ?speaker ?speakerLabel WHERE {
  VALUES ?language { entity:Q209 }
  ?speaker prop:P2 entity:Q3 .  # P2 'instance of' is Q3 'speaker'
  ?speaker prop:P4 ?language .  # P4 'language' is Q21 'French'
  # Labels
  SERVICE wikibase:label {
    bd:serviceParam wikibase:language "[AUTO_LANGUAGE],en" .
  } 
}
... Loading ...

✅ Language LL Qid (Q21) + Speaker (Q42) → Count records

SELECT ?language ?speakerLabel (COUNT(?audio) AS ?audio)
WHERE {
  VALUES ?language { entity:Q21 }
  VALUES ?speaker { entity:Q42 }
  ?audio prop:P4 ?language .  # P4 'language' is Q21 'French'
  ?audio prop:P2 entity:Q2 .  # P2 'instance of' is Q2 'record'
  ?audio prop:P5 ?speaker . # P5 'speaker' is Q42 '0x010C'
  # Labels
  SERVICE wikibase:label {bd:serviceParam wikibase:language "[AUTO_LANGUAGE],en"} 
}
GROUP BY ?language ?speakerLabel
... Loading ...

✅ Languages → Languages iso-639-3

SELECT * WHERE {
  ?lang prop:P13 ?code .
}
... Loading ...

✅ Isolang → Language LL Qid

SELECT ?langIso ?langId
WHERE {
  VALUES ?langIso { "ban" "bre" } # One or multiple values
  # P2 'instance of'; Q4 'language'; P13 'ISO 639-3 code'
  ?langId prop:P2 entity:Q4 ; prop:P13 ?langIso .
}
... Loading ...

Isolang → Language WD Qid

✅ Language WD Qid → Language data, all

SELECT * WHERE {
  ?lang prop:P12 "Q12107" .  # P12 'Wikidata id' is Wikidata's "Q12107"
  ?lang ?predicate ?object . # 
}
... Loading ...

✅ Language LL Qid (Q209) → Language data, all

'Case: Get for language Q209 'Breton' all its data.

SELECT * WHERE {
  # Given Q209 'Breton language', get all properties and values
  entity:Q209 ?predicate ?object .
}
... Loading ...

✅ Language LL Qid (Q209) → Language data, core

'Case: Get for language Q209 'Breton' all its CORE data.

SELECT * WHERE {
  # Given Q209 'Breton language', get all properties and values
  entity:Q209 ?predicate ?object .
  ?predicate rdf:type owl:DatatypeProperty .
}
... Loading ...

Records

✅ Item name + language → Qid(s)

SELECT ?itemLabel ?item
WHERE { 
  ?item prop:P2 entity:Q2 .    # Filter: P2 'instance of' Q3 'record'
  ?item rdfs:label ?itemLabel. # Assign value: label to ?itemLabel
  FILTER(CONTAINS(?itemLabel, "apple"@en)). 
} limit 10
... Loading ...

Audio Qid → Audio data

✅ Language (Q209) + speaker (Q584098) + word (ni) → Audio's Qid

Case: Search in Breton language, with speaker 'ThonyVezbe',

SELECT ?audio
WHERE {
  ?audio prop:P4 entity:Q209 .    # P4 'language' is Q209 'Breton'
  ?audio prop:P5 entity:Q584098 . # P5 'speaker' is Q584098 'ThonyVezbe'
  ?audio rdfs:label ?word . #word
  FILTER ( STR(?word) = "ni" )    # word = 'ni'
}
... Loading ...

Langue + speaker + word → Audio's Commons url pointeur (P3)

SELECT ?word ?audio ?url (STR(?url) AS ?urlStr)
WHERE {
  ?audio prop:P4 entity:Q209 .    # P4 'language' is Q209 'Breton'
  ?audio prop:P5 entity:Q584098 . # P5 'speaker' is Q584098 'ThonyVezbe'
  ?audio rdfs:label ?word .       # word
  FILTER ( STR(?word) = "achantour" )    # Filter: word is 'achantour'
  ?audio prop:P3 ?url
}
... Loading ...

Heavy queries

Queries below are too large to run on LinguaLibre's wikipages, or even on Lingualibre Query Service).
To do: do smaller sub-queries, with one COUNT() function.

❌ Languages → Name, Wikidata Qid, LLQid, Iso-639-3, and genders

Query Result
SELECT ?languageQidLabel ?wdQid ?languageQid ?isoCode 
(COUNT(DISTINCT(?record)) AS ?recordCount)
(COUNT(DISTINCT(?speakerLangM)) AS ?speakerM) 
(COUNT(DISTINCT(?speakerLangF)) AS ?speakerF)
wWHERE{
  ?record prop:P2 entity:Q2 .     # Filter: items where P2 'instance of' is Q2 'record'
  ?record prop:P4 ?languageQid .  # Assign value: P4 'language' into variable ?language
  ?languageQid prop:P12 ?wdQid .  # Assign value: P12 'wikidata id' into variable ?WD
  ?languageQid prop:P13 ?isoCode. # Assign value: P13 'iso639-3' into ?isoCode
  
  #?record prop:P5 ?speakerQidM .   # Assign value: P5 'speaker' into variable ?speakerQidM
  #?speakerQidM prop:P8 entity:Q16 .   # Filter: P8 'sex or gender' is Q16 'male
  #?speakerQidM prop:P4 ?speakerLangM .  # Assign value: P4 'language' into variable ?spakerLangM
  
  ?record prop:P5 ?speakerQidF .   # Assign value: P5 'speaker' into variable ?speakerQidF
  ?speakerQidF prop:P8 entity:Q17 .   # Filter: P8 'sex or gender' is Q17 'female
  ?speakerQidF prop:P4 ?speakerLangF .  # Assign value: P4 'language' into variable ?spakerLangF
  
  SERVICE wikibase:label { bd:serviceParam wikibase:language "en" . } 
}
GROUP BY ?languageQidLabel ?languageQid ?wdQid ?isoCode
ORDER BY DESC(?recordCount)
languageQidLabel	wdQid	languageQid	isoCode	recordCount	speakerM	speakerF
French	Q150	Q21	fra	16761	0	18
Marathi	Q1571	Q34	mar	13153	0	5
Polish	Q809	Q298	pol	11686	0	1
…

❌ Is Language (Q3) → list all languages with number of unique words and speakers

SELECT ?language (COUNT(?audio) AS ?nbAudio) (COUNT(?speaker) AS ?nbSpeaker) WHERE {
  ?language prop:P2 entity:Q4 .
  ?audio prop:P4 ?language .
  ?speaker prop:P4 ?language .
}
GROUP BY ?language

Tools

Wikidata lexemes

You may [<tvar|hackme>https://jsfiddle.net/hugolpz/rygo9s5b/</> Hack me !]

It is possible to use Wikidata & Wiktionary to extract lexicographical information. Developer <tvar|sina_ahm>@sina_ahm</> created a SPARQL query generator, helping us to search words in both Wikidata & Dbnary. See demo there : <tvar|1>https://sinaahmadi.github.io/posts/sparql-query-generator-for-lexicographical-data.html</>

See also