#124 Fixes for update_data.py and reformat/add to noun queries

scribe-org · Jun 2, 2024 · bff1da6 · bff1da6
1 parent 22632f5
commit bff1da6
Show file tree

Hide file tree

Showing 87 changed files with 605 additions and 483 deletions.
diff --git a/src/scribe_data/language_data_extraction/Arabic/nouns/query_nouns.sparql b/src/scribe_data/language_data_extraction/Arabic/nouns/query_nouns.sparql
@@ -1,17 +1,35 @@
 # tool: scribe-data
-# All Arabic (Q13955) nouns.
+# All Arabic (Q13955) nouns, their plurals and their dual forms.
 # Enter this query at https://query.wikidata.org/.
 
 SELECT DISTINCT
-  (REPLACE(STR(?lexeme), "http://www.wikidata.org/entity/", "") as ?lexemeID)
-  ?noun
+  (REPLACE(STR(?lexeme), "http://www.wikidata.org/entity/", "") AS ?lexemeID)
+  ?singular
+  ?plural
+  ?dual
 
 WHERE {
-  # Nouns and pronouns.
-  VALUES ?nounTypes { wd:Q1084 wd:Q147276 }
+  VALUES ?nounTypes {wd:Q1084 wd:Q147276} # Nouns and pronouns # Nouns and pronouns
   ?lexeme a ontolex:LexicalEntry ;
     dct:language wd:Q13955 ;
     wikibase:lexicalCategory ?nounType ;
-    wikibase:lemma ?noun .
+    wikibase:lemma ?singular .
+
   FILTER(?nounType = ?nounTypes)
+
+  # Optional selection of indefinite plural forms.
+  OPTIONAL {
+    ?lexeme ontolex:lexicalForm ?pluralForm .
+    ?pluralForm ontolex:representation ?plural ;
+      wikibase:grammaticalFeature wd:Q53997857 ;
+      wikibase:grammaticalFeature wd:Q146786 ;
+  } .
+
+  # Optional selection of indefinite dual form.
+  OPTIONAL {
+    ?lexeme ontolex:lexicalForm ?dualForm .
+    ?dualForm ontolex:representation ?dual ;
+      wikibase:grammaticalFeature wd:Q53997857 ;
+      wikibase:grammaticalFeature wd:Q110022 ;
+  } .
 }
diff --git a/src/scribe_data/language_data_extraction/Arabic/verbs/query_verbs_1.sparql b/src/scribe_data/language_data_extraction/Arabic/verbs/query_verbs_1.sparql
@@ -3,7 +3,7 @@
 # Enter this query at https://query.wikidata.org/.
 
 SELECT
-  (REPLACE(STR(?lexeme), "http://www.wikidata.org/entity/", "") as ?lexemeID)
+  (REPLACE(STR(?lexeme), "http://www.wikidata.org/entity/", "") AS ?lexemeID)
   ?verb
   ?presFPS ?presSPSM ?presSPSF ?presTPSM ?presTPSF
   ?presSPD ?presTPDM ?presTPDF

diff --git a/src/scribe_data/language_data_extraction/Arabic/verbs/query_verbs_2.sparql b/src/scribe_data/language_data_extraction/Arabic/verbs/query_verbs_2.sparql
@@ -3,7 +3,7 @@
 # Enter this query at https://query.wikidata.org/.
 
 SELECT
-  (REPLACE(STR(?lexeme), "http://www.wikidata.org/entity/", "") as ?lexemeID)
+  (REPLACE(STR(?lexeme), "http://www.wikidata.org/entity/", "") AS ?lexemeID)
   ?verb
   ?pastFPS ?pastSPSM ?pastSPSF ?pastTPSM ?pastTPSF
   ?pastSPD ?pastTPDM ?pastTPDF

diff --git a/src/scribe_data/language_data_extraction/Arabic/verbs/query_verbs_3.sparql b/src/scribe_data/language_data_extraction/Arabic/verbs/query_verbs_3.sparql
@@ -3,7 +3,7 @@
 # Enter this query at https://query.wikidata.org/.
 
 SELECT
-  (REPLACE(STR(?lexeme), "http://www.wikidata.org/entity/", "") as ?lexemeID)
+  (REPLACE(STR(?lexeme), "http://www.wikidata.org/entity/", "") AS ?lexemeID)
   ?verb
   ?impSPSM ?impSPSF ?impSPD
   ?impSPPM ?impSPPF

diff --git a/src/scribe_data/language_data_extraction/Basque/nouns/query_nouns.sparql b/src/scribe_data/language_data_extraction/Basque/nouns/query_nouns.sparql
@@ -1,17 +1,35 @@
 # tool: scribe-data
-# All Basque (Q8752) nouns.
+# All Basque (Q8752) nouns and all implemented singular and plural forms.
 # Enter this query at https://query.wikidata.org/.
 
 SELECT DISTINCT
-  (REPLACE(STR(?lexeme), "http://www.wikidata.org/entity/", "") as ?lexemeID)
-  ?noun
+  (REPLACE(STR(?lexeme), "http://www.wikidata.org/entity/", "") AS ?lexemeID)
+  ?absIndefinite
+  ?absSingular
+  ?absPlural
 
 WHERE {
-  # Nouns and pronouns.
-  VALUES ?nounTypes { wd:Q1084 wd:Q147276 }
+  VALUES ?nounTypes {wd:Q1084 wd:Q147276} # Nouns and pronouns
   ?lexeme a ontolex:LexicalEntry ;
     dct:language wd:Q8752 ;
     wikibase:lexicalCategory ?nounType ;
-    wikibase:lemma ?noun .
+    wikibase:lemma ?absIndefinite .
+
   FILTER(?nounType = ?nounTypes)
+
+  # Optional selection of absolutive singular forms.
+  OPTIONAL {
+    ?lexeme ontolex:lexicalForm ?absSingularForm .
+    ?absSingularForm ontolex:representation ?absSingular ;
+      wikibase:grammaticalFeature wd:Q332734 ;
+      wikibase:grammaticalFeature wd:Q110786 ;
+  } .
+
+  # Optional selection of absolutive plural forms.
+  OPTIONAL {
+    ?lexeme ontolex:lexicalForm ?absPluralForm .
+    ?absPluralForm ontolex:representation ?absPlural ;
+      wikibase:grammaticalFeature wd:Q332734 ;
+      wikibase:grammaticalFeature wd:Q146786 ;
+  } .
 }
diff --git a/src/scribe_data/language_data_extraction/Basque/verbs/query_verbs.sparql b/src/scribe_data/language_data_extraction/Basque/verbs/query_verbs.sparql
@@ -3,7 +3,7 @@
 # Enter this query at https://query.wikidata.org/.
 
 SELECT
-  (REPLACE(STR(?lexeme), "http://www.wikidata.org/entity/", "") as ?lexemeID)
+  (REPLACE(STR(?lexeme), "http://www.wikidata.org/entity/", "") AS ?lexemeID)
   ?verb
 
 WHERE {

diff --git a/src/scribe_data/language_data_extraction/Bengali/adjectives/query_adjectives.sparql b/src/scribe_data/language_data_extraction/Bengali/adjectives/query_adjectives.sparql
@@ -3,7 +3,7 @@
 # Enter this query at https://query.wikidata.org/.
 
 SELECT
-  (REPLACE(STR(?lexeme), "http://www.wikidata.org/entity/", "") as ?lexemeID)
+  (REPLACE(STR(?lexeme), "http://www.wikidata.org/entity/", "") AS ?lexemeID)
   ?adjective
 
 WHERE {

diff --git a/src/scribe_data/language_data_extraction/Bengali/nouns/query_nouns.sparql b/src/scribe_data/language_data_extraction/Bengali/nouns/query_nouns.sparql
@@ -1,17 +1,47 @@
 # tool: scribe-data
-# All Bengali (Q9610) nouns.
+# All Bengali (Q9610) nouns and their forms in the various cases.
 # Enter this query at https://query.wikidata.org/.
 
 SELECT DISTINCT
-  (REPLACE(STR(?lexeme), "http://www.wikidata.org/entity/", "") as ?lexemeID)
-  ?noun
+  (REPLACE(STR(?lexeme), "http://www.wikidata.org/entity/", "") AS ?lexemeID)
+  ?nominative
+  ?genitive
+  ?accusative
+  ?locative
 
 WHERE {
-  # Nouns and pronouns.
-  VALUES ?nounTypes { wd:Q1084 wd:Q147276 }
+  VALUES ?nounTypes {wd:Q1084 wd:Q147276} # Nouns and pronouns
   ?lexeme a ontolex:LexicalEntry ;
     dct:language wd:Q9610 ;
     wikibase:lexicalCategory ?nounType ;
-    wikibase:lemma ?noun .
+
   FILTER(?nounType = ?nounTypes)
+
+  # Optional selection of nominative form.
+  OPTIONAL {
+    ?lexeme ontolex:lexicalForm ?nomForm .
+    ?nomForm ontolex:representation ?nominative ;
+      wikibase:grammaticalFeature wd:Q131105 ;
+  } .
+
+  # Optional selection of genitive form.
+  OPTIONAL {
+    ?lexeme ontolex:lexicalForm ?genForm .
+    ?genForm ontolex:representation ?genitive ;
+      wikibase:grammaticalFeature wd:Q146233 ;
+  } .
+
+  # Optional selection of accusative form.
+  OPTIONAL {
+    ?lexeme ontolex:lexicalForm ?accForm .
+    ?accForm ontolex:representation ?accusative ;
+      wikibase:grammaticalFeature wd:Q146078 ;
+  } .
+
+  # Optional selection of locative form.
+  OPTIONAL {
+    ?lexeme ontolex:lexicalForm ?locForm .
+    ?locForm ontolex:representation ?locative ;
+      wikibase:grammaticalFeature wd:Q202142 ;
+  } .
 }
diff --git a/src/scribe_data/language_data_extraction/Bengali/prepositions/query_prepositions.sparql b/src/scribe_data/language_data_extraction/Bengali/prepositions/query_prepositions.sparql
@@ -3,7 +3,7 @@
 # Enter this query at https://query.wikidata.org/.
 
 SELECT
-  (REPLACE(STR(?lexeme), "http://www.wikidata.org/entity/", "") as ?lexemeID)
+  (REPLACE(STR(?lexeme), "http://www.wikidata.org/entity/", "") AS ?lexemeID)
   ?preposition
   ?case
 

diff --git a/src/scribe_data/language_data_extraction/Bengali/verbs/query_verbs.sparql b/src/scribe_data/language_data_extraction/Bengali/verbs/query_verbs.sparql
@@ -3,7 +3,7 @@
 # Enter this query at https://query.wikidata.org/.
 
 SELECT
-  (REPLACE(STR(?lexeme), "http://www.wikidata.org/entity/", "") as ?lexemeID)
+  (REPLACE(STR(?lexeme), "http://www.wikidata.org/entity/", "") AS ?lexemeID)
   ?verb
 
 WHERE {

diff --git a/src/scribe_data/language_data_extraction/Bokmål/nouns/query_nouns.sparql b/src/scribe_data/language_data_extraction/Bokmål/nouns/query_nouns.sparql
@@ -1,41 +1,55 @@
 # tool: scribe-data
-# All Bokmål (Norwegian) (Q9043) nouns, their plural and their gender.
+# All Bokmål Norwegian (Q9043) nouns, their plurals and their genders.
 # Enter this query at https://query.wikidata.org/.
 # Note that this query is for Bokmål (Q25167) rather than Nynorsk (Q25164).
 
 SELECT DISTINCT
-  (REPLACE(STR(?lexeme), "http://www.wikidata.org/entity/", "") as ?lexemeID)
-  ?singular
-  ?plural
+  (REPLACE(STR(?lexeme), "http://www.wikidata.org/entity/", "") AS ?lexemeID)
+  ?indefSingular
+  ?defSingular
+  ?indefPlural
+  ?defPlural
   ?gender
 
 WHERE {
-
-  # Nouns and pronouns.
-  VALUES ?nounTypes { wd:Q1084 wd:Q147276 }
+  VALUES ?nounTypes {wd:Q1084 wd:Q147276} # Nouns and pronouns
   ?lexeme a ontolex:LexicalEntry ;
     dct:language wd:Q25167 ;
-    wikibase:lexicalCategory ?noun .
-  FILTER(?noun = ?nounTypes)
+    wikibase:lexicalCategory ?nounType ;
+    wikibase:lemma ?indefSingular .
+
+  FILTER(?nounType = ?nounTypes)
 
-  # Optional selection of singular forms.
+  # Optional selection of definite singular forms.
   OPTIONAL {
-    ?lexeme ontolex:lexicalForm ?singularForm .
-    ?singularForm ontolex:representation ?singular ;
+    ?lexeme ontolex:lexicalForm ?  ?defSingularForm .
+    ?defSingularForm ontolex:representation ?defSingular ;
       wikibase:grammaticalFeature wd:Q110786 ;
+      wikibase:grammaticalFeature wd:Q53997851 ;
+  } .
+
+  # Optional selection of indefinite plural forms.
+  OPTIONAL {
+    ?lexeme ontolex:lexicalForm ?indefPluralForm .
+    ?indefPluralForm ontolex:representation ?indefPlural ;
+      wikibase:grammaticalFeature wd:Q146786 ;
+      wikibase:grammaticalFeature wd:Q53997857 ;
   } .
 
-  # Optional selection of plural forms.
+  # Optional selection of definite plural forms.
   OPTIONAL {
-    ?lexeme ontolex:lexicalForm ?pluralForm .
-    ?pluralForm ontolex:representation ?plural ;
+    ?lexeme ontolex:lexicalForm ?defPluralForm .
+    ?defPluralForm ontolex:representation ?defPlural ;
       wikibase:grammaticalFeature wd:Q146786 ;
+      wikibase:grammaticalFeature wd:Q53997851 ;
   } .
 
   # Optional selection of genders.
   OPTIONAL {
     ?lexeme wdt:P5185 ?nounGender .
-    FILTER NOT EXISTS { ?lexeme wdt:P31 wd:Q202444 . }
+    FILTER NOT EXISTS {
+      ?lexeme wdt:P31 wd:Q202444 .
+    }
   } .
 
   SERVICE wikibase:label {

diff --git a/src/scribe_data/language_data_extraction/Bokmål/verbs/query_verbs.sparql b/src/scribe_data/language_data_extraction/Bokmål/verbs/query_verbs.sparql
@@ -4,7 +4,7 @@
 # Note that this query is for Bokmål (Q25167) rather than Nynorsk (Q25164).
 
 SELECT DISTINCT
-  (REPLACE(STR(?lexeme), "http://www.wikidata.org/entity/", "") as ?lexemeID)
+  (REPLACE(STR(?lexeme), "http://www.wikidata.org/entity/", "") AS ?lexemeID)
   ?infinitive
 
 WHERE {

diff --git a/src/scribe_data/language_data_extraction/Czech/nouns/query_nouns.sparql b/src/scribe_data/language_data_extraction/Czech/nouns/query_nouns.sparql
@@ -1,41 +1,43 @@
 # tool: scribe-data
-# All Czeck (Q9056) nouns, their plural, and their gender.
+# All Czeck (Q9056) nouns, their plurals and their genders.
 # Enter this query at https://query.wikidata.org/.
 
 SELECT DISTINCT
-  (REPLACE(STR(?lexeme), "http://www.wikidata.org/entity/", "") as ?lexemeID)
-  ?singular
-  ?plural
+  (REPLACE(STR(?lexeme), "http://www.wikidata.org/entity/", "") AS ?lexemeID)
+  ?nomSingular
+  ?nomPlural
   ?gender
 
 WHERE {
-  # Nouns and pronouns.
-  VALUES ?nounTypes { wd:Q1084 wd:Q147276 }
+  VALUES ?nounTypes {wd:Q1084 wd:Q147276} # Nouns and pronouns
   ?lexeme a ontolex:LexicalEntry ;
     dct:language wd:Q9056 ;
-    wikibase:lexicalCategory ?noun .
-  FILTER(?noun = ?nounTypes)
+    wikibase:lexicalCategory ?nounType .
 
-  # Optional selection of singular forms.
+  FILTER(?nounType = ?nounTypes)
+
+  # Optional selection of nominative singular forms.
   OPTIONAL {
-    ?lexeme ontolex:lexicalForm ?singularForm .
-    ?singularForm ontolex:representation ?singular ;
-      wikibase:grammaticalFeature wd:Q110786 ;
+    ?lexeme ontolex:lexicalForm ?nomSingularForm .
+    ?nomSingularForm ontolex:representation ?nomSingular ;
       wikibase:grammaticalFeature wd:Q131105 ;
+      wikibase:grammaticalFeature wd:Q110786 ;
   } .
 
-  # Optional selection of plural forms.
+  # Optional selection of nominative plural forms.
   OPTIONAL {
-    ?lexeme ontolex:lexicalForm ?pluralForm .
-    ?pluralForm ontolex:representation ?plural ;
-      wikibase:grammaticalFeature wd:Q146786 ;
+    ?lexeme ontolex:lexicalForm ?nomPluralForm .
+    ?nomPluralForm ontolex:representation ?nomPlural ;
       wikibase:grammaticalFeature wd:Q131105 ;
+      wikibase:grammaticalFeature wd:Q146786 ;
   } .
 
   # Optional selection of genders.
   OPTIONAL {
     ?lexeme wdt:P5185 ?nounGender .
-    FILTER NOT EXISTS { ?lexeme wdt:P31 wd:Q202444 . }
+    FILTER NOT EXISTS {
+      ?lexeme wdt:P31 wd:Q202444 .
+    }
   } .
 
   SERVICE wikibase:label {

diff --git a/src/scribe_data/language_data_extraction/Czech/prepositions/query_prepositions.sparql b/src/scribe_data/language_data_extraction/Czech/prepositions/query_prepositions.sparql
@@ -3,7 +3,7 @@
 # Enter this query at https://query.wikidata.org/.
 
 SELECT
-  (REPLACE(STR(?lexeme), "http://www.wikidata.org/entity/", "") as ?lexemeID)
+  (REPLACE(STR(?lexeme), "http://www.wikidata.org/entity/", "") AS ?lexemeID)
   ?preposition
   ?case
 

diff --git a/src/scribe_data/language_data_extraction/Czech/verbs/query_verbs.sparql b/src/scribe_data/language_data_extraction/Czech/verbs/query_verbs.sparql
@@ -3,7 +3,7 @@
 # Enter this query at https://query.wikidata.org/.
 
 SELECT
-  (REPLACE(STR(?lexeme), "http://www.wikidata.org/entity/", "") as ?lexemeID)
+  (REPLACE(STR(?lexeme), "http://www.wikidata.org/entity/", "") AS ?lexemeID)
   ?infinitive
 
 WHERE {