Merge pull request #1097 from globalwordnet/issue-1044

Make sure that no instances have hyponyms

scripts/sense_keys.py

@@ -4,53 +4,53 @@
 from sys import exit
 
 lex_filenums = {
-    "src/xml/wn-adj.all.xml": 0,
-    "src/xml/wn-adj.pert.xml": 1,
-    "src/xml/wn-adv.all.xml": 2,
-    "src/xml/wn-noun.Tops.xml": 3,
-    "src/xml/wn-noun.act.xml": 4,
-    "src/xml/wn-noun.animal.xml": 5,
-    "src/xml/wn-noun.artifact.xml": 6,
-    "src/xml/wn-noun.attribute.xml": 7,
-    "src/xml/wn-noun.body.xml": 8,
-    "src/xml/wn-noun.cognition.xml": 9,
-    "src/xml/wn-noun.communication.xml": 10,
-    "src/xml/wn-noun.event.xml": 11,
-    "src/xml/wn-noun.feeling.xml": 12,
-    "src/xml/wn-noun.food.xml": 13,
-    "src/xml/wn-noun.group.xml": 14,
-    "src/xml/wn-noun.location.xml": 15,
-    "src/xml/wn-noun.motive.xml": 16,
-    "src/xml/wn-noun.object.xml": 17,
-    "src/xml/wn-noun.person.xml": 18,
-    "src/xml/wn-noun.phenomenon.xml": 19,
-    "src/xml/wn-noun.plant.xml": 20,
-    "src/xml/wn-noun.possession.xml": 21,
-    "src/xml/wn-noun.process.xml": 22,
-    "src/xml/wn-noun.quantity.xml": 23,
-    "src/xml/wn-noun.relation.xml": 24,
-    "src/xml/wn-noun.shape.xml": 25,
-    "src/xml/wn-noun.state.xml": 26,
-    "src/xml/wn-noun.substance.xml": 27,
-    "src/xml/wn-noun.time.xml": 28,
-    "src/xml/wn-verb.body.xml": 29,
-    "src/xml/wn-verb.change.xml": 30,
-    "src/xml/wn-verb.cognition.xml": 31,
-    "src/xml/wn-verb.communication.xml": 32,
-    "src/xml/wn-verb.competition.xml": 33,
-    "src/xml/wn-verb.consumption.xml": 34,
-    "src/xml/wn-verb.contact.xml": 35,
-    "src/xml/wn-verb.creation.xml": 36,
-    "src/xml/wn-verb.emotion.xml": 37,
-    "src/xml/wn-verb.motion.xml": 38,
-    "src/xml/wn-verb.perception.xml": 39,
-    "src/xml/wn-verb.possession.xml": 40,
-    "src/xml/wn-verb.social.xml": 41,
-    "src/xml/wn-verb.stative.xml": 42,
-    "src/xml/wn-verb.weather.xml": 43,
-    "src/xml/wn-adj.ppl.xml": 44,
-    "src/xml/wn-contrib.colloq.xml": 50,
-    "src/xml/wn-contrib.plwn.xml": 51}
+    "adj.all": 0,
+    "adj.pert": 1,
+    "adv.all": 2,
+    "noun.Tops": 3,
+    "noun.act": 4,
+    "noun.animal": 5,
+    "noun.artifact": 6,
+    "noun.attribute": 7,
+    "noun.body": 8,
+    "noun.cognition": 9,
+    "noun.communication": 10,
+    "noun.event": 11,
+    "noun.feeling": 12,
+    "noun.food": 13,
+    "noun.group": 14,
+    "noun.location": 15,
+    "noun.motive": 16,
+    "noun.object": 17,
+    "noun.person": 18,
+    "noun.phenomenon": 19,
+    "noun.plant": 20,
+    "noun.possession": 21,
+    "noun.process": 22,
+    "noun.quantity": 23,
+    "noun.relation": 24,
+    "noun.shape": 25,
+    "noun.state": 26,
+    "noun.substance": 27,
+    "noun.time": 28,
+    "verb.body": 29,
+    "verb.change": 30,
+    "verb.cognition": 31,
+    "verb.communication": 32,
+    "verb.competition": 33,
+    "verb.consumption": 34,
+    "verb.contact": 35,
+    "verb.creation": 36,
+    "verb.emotion": 37,
+    "verb.motion": 38,
+    "verb.perception": 39,
+    "verb.possession": 40,
+    "verb.social": 41,
+    "verb.stative": 42,
+    "verb.weather": 43,
+    "adj.ppl": 44,
+    "contrib.colloq": 50,
+    "contrib.plwn": 51}
 
 ss_types = {
     PartOfSpeech.NOUN: 1,
@@ -118,16 +118,14 @@ def get_head_word(wn, s):
     exit(-1)
 
 
-def get_sense_key(wn, e, s, wn_file):
+def get_sense_key(wn, e, s):
     """Calculate the sense key for a sense of an entry"""
     ss = wn.synset_by_id(s.synset)
     lemma = e.lemma.written_form.replace(
         " ", "_").replace(
         "&apos", "'").lower()
     ss_type = ss_types[ss.part_of_speech]
-    if not wn_file.startswith("src/xml/wn-"):
-        wn_file = f"src/xml/wn-{wn_file}.xml"
-    lex_filenum = lex_filenums[wn_file]
+    lex_filenum = lex_filenums[ss.lex_name]
     if s.id:
         lex_id = extract_lex_id(unmap_sense_key(s.id))
     else:

scripts/validate.py

@@ -141,40 +141,29 @@ def check_lex_files(wn, fix):
         "adv": PartOfSpeech.ADVERB
     }
     errors = 0
-    for f in glob.glob("src/xml/wn-*.xml"):
-        lexfile = f[11:-4]
-        lex_pos = pos_map[lexfile[:3]]
-        swn = parse_wordnet(f)
-        for synset in swn.synsets:
-            if synset.lex_name != lexfile:
-                print("%s declared in %s but listed as %s" %
-                      (synset.id, lexfile, synset.lex_name))
+    for entry in wn.entries:
+        for sense in entry.senses:
+            if not sense.id:
+                print("%s does not have a sense key" % (sense.id))
                 errors += 1
-            if not equal_pos(lex_pos, synset.part_of_speech):
-                print("%s declared in %s but has wrong POS %s" %
-                      (synset.id, lexfile, synset.part_of_speech))
+            if not wn.synset_by_id(sense.synset):
+                print("%s refers to nonexistent synset %s" %
+                      (sense.id, sense.synset))
                 errors += 1
-        for entry in swn.entries:
-            if len(entry.senses) == 0:
-                print("%s is empty in %s" % (entry.id, lexfile))
+                continue
+            calc_sense_key = sense_keys.get_sense_key(
+                wn, entry, sense)
+            sense_key = unmap_sense_key(sense.id)
+            if sense_key != calc_sense_key:
+                if fix:
+                    print(
+                        "sed -i 's/%s/%s/' src/xml/*" %
+                        (sense_key, calc_sense_key))
+                else:
+                    print(
+                        "%s has declared key %s but should be %s" %
+                        (sense.id, sense_key, calc_sense_key))
                 errors += 1
-            for sense in entry.senses:
-                if not sense.id:
-                    print("%s does not have a sense key" % (sense.id))
-                    errors += 1
-                calc_sense_key = sense_keys.get_sense_key(
-                    wn, entry, sense, f)
-                sense_key = unmap_sense_key(sense.id)
-                if sense_key != calc_sense_key:
-                    if fix:
-                        print(
-                            "sed -i 's/%s/%s/' src/xml/*" %
-                            (sense_key, calc_sense_key))
-                    else:
-                        print(
-                            "%s has declared key %s but should be %s" %
-                            (sense.id, sense_key, calc_sense_key))
-                    errors += 1
 
     return errors
 
@@ -286,6 +275,8 @@ def main():
                         sense.id, sense2.id, sense.synset))
                     errors += 1
 
+    instances = set()
+
     for synset in wn.synsets:
         if synset.id[-1:] != synset.part_of_speech.value:
             print(
@@ -356,6 +347,15 @@ def main():
             print("ERROR: noun synset %s has no hypernym" % synset.id)
             errors += 1
 
+        if any(sr.rel_type == SynsetRelType.INSTANCE_HYPERNYM 
+               for sr in synset.synset_relations):
+            if any(sr.rel_type == SynsetRelType.HYPERNYM
+                   for sr in synset.synset_relations):
+                print("Error: synset %s has both hypernym and instance hypernym"
+                      % synset.id)
+                errors += 1
+            instances.add(synset.id)
+
         if len(synset.definitions) == 0:
             print("ERROR: synset without definition %s" % (synset.id))
             errors += 1
@@ -373,6 +373,15 @@ def main():
                     (synset.id, item[1], item[0]))
                 errors += 1
 
+    for synset in wn.synsets:
+        for sr in synset.synset_relations:
+            if sr.rel_type == SynsetRelType.HYPERNYM:
+                if sr.target in instances:
+                    print(
+                        "ERROR: Hypernym targets instance %s => %s" %
+                        (synset.id, sr.target))
+                    errors += 1
+
     for error in check_symmetry(wn, fix):
         if fix:
             print(error)

src/deprecations.csv

@@ -119,6 +119,10 @@
 "ewn-01798679-v","i30704","ewn-00064841-v","i22102","Duplicate (#1043)"
 "ewn-02125976-v","i32364","ewn-00064841-v","i22102","Duplicate (#1043)"
 "ewn-00078154-v","i22163","ewn-00064841-v","i22102","Duplicate (#1043)"
+"ewn-09369905-n","i85747","ewn-09372942-n","i85763","Duplicate (#1044)"
+"ewn-08893374-n","i83377","ewn-08879115-n","i83373","Merged with hypernym (#1044)"
+"ewn-90010441-n","","ewn-09067337-n","i84182","Merge with hypernym (#1044)"
+"ewn-00596900-v","i24739","ewn-01311849-s","i7131","Not attested (#1060)"
 "ewn-10809460-n","i94174","ewn-10146463-n","i90187","Compositional; use 10146463-n instead (#1068)"
 "ewn-83000076-n","","ewn-10146463-n","i90187","Compositional (#1068)"
 "ewn-10809055-n","i94171","ewn-10146463-n","i90187","Compositional (#1068)"

src/yaml/entries-b.yaml

@@ -6259,11 +6259,11 @@ Blighia sapida:
       synset: 12763050-n
 Blighty:
   n:
-    pronunciation:
-    - value: ˈblaɪti
     sense:
-    - id: 'blighty%1:15:00::'
-      synset: 08893374-n
+    - exemplifies:
+      - 'slang%1:10:01::'
+      id: 'blighty%1:15:01::'
+      synset: 08879115-n
 Blimp:
   n:
     pronunciation:

src/yaml/entries-m.yaml

@@ -24334,8 +24334,8 @@ mare liberum:
 mare nostrum:
   n:
     sense:
-    - id: 'mare_nostrum%1:17:00::'
-      synset: 09369905-n
+    - id: 'mare_nostrum%1:17:01::'
+      synset: 09372942-n
 mare's nest:
   n:
     pronunciation:
@@ -36913,8 +36913,10 @@ merging:
 merica:
   n:
     sense:
-    - id: 'merica%1:15:01::'
-      synset: 90010441-n
+    - exemplifies:
+      - 'slang%1:10:01::'
+      id: 'merica%1:15:01::'
+      synset: 09067337-n
 mericarp:
   n:
     sense:

src/yaml/noun.group.yaml

@@ -29276,11 +29276,10 @@
     and individualism instead of tradition and established doctrine
   example:
   - the Enlightenment brought about many humanitarian reforms
-  hypernym:
-  - 08490634-n
   ili: i81574
   instance_hypernym:
   - 15278839-n
+  - 08490634-n
   members:
   - Age of Enlightenment
   - Enlightenment