Add is_astronomical_object feature for wikidatawiki

articlequality/feature_lists/wikidatawiki.py

@@ -7,7 +7,8 @@
 from revscoring.features.meta import aggregators
 from revscoring.features.modifiers import not_
 
-from . import wikibase, property_datatypes
+from . import wikibase
+from .wikidatawiki_data import property_datatypes, items_lists
 
 name = "wikidatawiki"
 
@@ -139,6 +140,14 @@ def _process_important_description_translations(item_descriptions):
 "`int` : A count of unique sources in the revision"
 
 
+def _process_is_astronomical_object(entity):
+    statements = entity.properties.get(properties.INSTANCE_OF, [])
+    for s in statements:
+        if str(s.claim.datavalue) in items_lists.ASTRONOMICAL_OBJECTS:
+            return True
+    return False
+
+
 def _process_item_completeness(current_properties, properties_suggested):
     current_properties = set(current_properties.keys())
 
@@ -173,10 +182,15 @@ def _process_item_completeness(current_properties, properties_suggested):
     items.SCHOLARLY_ARTICLE,
     name=name + '.revision.is_scholarlyarticle'
 )
-
+is_astronomical_object = Feature(
+    name + '.revision.page.is_astronomical_object',
+    _process_is_astronomical_object,
+    returns=bool,
+    depends_on=[wikibase_.revision.datasources.entity])
 
 local_wiki = [
     is_scholarlyarticle,
+    is_astronomical_object,
     is_human,
     is_blp,
     aggregators.len(complete_translations),

articlequality/feature_lists/wikidatawiki_data/items_lists.py

@@ -0,0 +1,12 @@
+# The list is generated using https://w.wiki/aaD
+# Picking the top 50 ones only because it covers 97% of cases
+ASTRONOMICAL_OBJECTS = [
+    'Q523', 'Q318', 'Q1931185', 'Q1457376', 'Q2247863', 'Q3863', 'Q83373',
+    'Q2154519', 'Q726242', 'Q1153690', 'Q204107', 'Q71963409', 'Q67206691',
+    'Q1151284', 'Q67206701', 'Q66619666', 'Q72802727', 'Q2168098', 'Q6243',
+    'Q72802508', 'Q11282', 'Q72803170', 'Q1332364', 'Q72802977', 'Q6999',
+    'Q1491746', 'Q272447', 'Q497654', 'Q204194', 'Q130019', 'Q744691',
+    'Q71798532', 'Q46587', 'Q11276', 'Q71965429', 'Q5871', 'Q72803622',
+    'Q72803426', 'Q3937', 'Q72803708', 'Q168845', 'Q24452', 'Q67201574',
+    'Q2557101', 'Q691269', 'Q13632', 'Q10451997', 'Q28738741', 'Q22247'
+]

articlequality/feature_lists/property_datatypes.py → articlequality/feature_lists/wikidatawiki_data/property_datatypes.py

@@ -1,14 +1,16 @@
 # Wikidata feature lists
 
-# To add/delete new properties from the `wikidatawiki.py` feature list we need to manually 
-# update this file. 
+# To add/delete new properties from the `wikidatawiki.py` feature list we need
+# to manually update this file.
 
-# Currently the file contains a list of `NON_EXTERNAL_IDENTIFIERS` that we want to extract 
-# as features. `NON_EXTERNAL_IDENTIFIERS` is used instead of `EXTERNAL_IDENTIFIERS` because 
-# external identifiers comprise about 5000+ of the total 8000+ currently on the database. 
-# Using NON_EXTERNAL_IDENTIFIERS (~2500) reduces the processing time. 
+# Currently the file contains a list of `NON_EXTERNAL_IDENTIFIERS` that we
+# want to extract as features. `NON_EXTERNAL_IDENTIFIERS` is used instead of
+# `EXTERNAL_IDENTIFIERS` because external identifiers comprise about 5000+ of
+# the total 8000+ currently on the database.
+# Using NON_EXTERNAL_IDENTIFIERS (~2500) reduces the processing time.
 
-# This list was obtained by making this query: https://quarry.wmflabs.org/query/47645
+# This list was obtained by making this query:
+# https://quarry.wmflabs.org/query/47645
 # and then doing a search and replace to format it as a python list.
 
 NONEXTERNAL_IDENTIFIERS = [

tests/feature_lists/test_wikidatawiki.py

@@ -57,3 +57,39 @@ def test_references_features(q7251):
 def test_external_identifiers(q7251):
     assert solve(aggregators.len(wikidatawiki.external_identifiers),
                  cache={entity: q7251}) == 79
+
+
+def test_is_astronomical_object(q7251):
+    crab_nebula = {
+        'title': 'Q207436',
+        'id': 'Q207436',
+        'claims': {
+            'P31': [
+                {
+                    'mainsnak': {
+                        'snaktype': 'value',
+                        'property': 'P31',
+                        'datavalue': {
+                            'value': {
+                                'entity-type': 'item',
+                                'numeric-id': 1931185,
+                                'id': 'Q1931185'
+                            },
+                            'type': 'wikibase-entityid'
+                        },
+                        'datatype': 'wikibase-item'
+                    },
+                    'type': 'statement',
+                    'id': 'Q10934$46D9E951-DD5A-4832-A2D2-DB0CE0425E0E',
+                    'rank': 'normal'
+                }
+            ]
+        }
+    }
+
+    crab_nebula = mwbase.Entity.from_json(crab_nebula)
+
+    assert solve(wikidatawiki.is_astronomical_object,
+                 cache={entity: crab_nebula}) is True
+    assert solve(wikidatawiki.is_astronomical_object,
+                 cache={entity: q7251}) is False