Merge pull request #1122 from globalwordnet/issue-1121

Remove duplicate ILI links

Author: jmccrae

scripts/validate.py

@@ -277,6 +277,7 @@ def main():
                     errors += 1
 
     instances = set()
+    ilis = set()
 
     for synset in wn.synsets:
         if synset.id[-1:] != synset.part_of_speech.value:
@@ -374,6 +375,12 @@ def main():
                     (synset.id, item[1], item[0]))
                 errors += 1
 
+        if synset.ili != "in" and synset.ili in ilis:
+            print(f"ERROR: ILI {synset.ili} is duplicated")
+            errors += 1
+        else:
+            ilis.add(synset.ili)
+
     for synset in wn.synsets:
         for sr in synset.synset_relations:
             if sr.rel_type == SynsetRelType.HYPERNYM:

scripts/wordnet.py

@@ -102,7 +102,7 @@ def change_sense_id(self, sense, new_id):
         self.sense2synset[new_id] = sense.synset
         self.id2sense[new_id] = sense
 
-    def to_xml(self, xml_file, part=True):
+    def to_xml(self, xml_file, part=False):
         xml_file.write("""<?xml version="1.0" encoding="UTF-8"?>\n""")
         if part:
             xml_file.write(

src/yaml/adj.all.yaml

@@ -2222,7 +2222,6 @@
   - 08215965-n
   example:
   - a newly activated unit
-  ili: i216
   members:
   - activated
   partOfSpeech: s
@@ -53180,7 +53179,6 @@
   - a scenic but devious route
   - a long and circuitous journey by train and boat
   - a roundabout route avoided rush-hour traffic
-  ili: i4234
   members:
   - devious
   - circuitous

src/yaml/noun.act.yaml

@@ -51679,7 +51679,6 @@
   - the mastery of fire was a huge advance in human technology
   hypernym:
   - 00950950-n
-  ili: i40396
   members:
   - technology
   partOfSpeech: n

src/yaml/verb.cognition.yaml

@@ -5002,7 +5002,6 @@
   - Her agent had booked her for several concerts in Tokyo
   hypernym:
   - 00680201-v
-  ili: i34688
   members:
   - book
   partOfSpeech: v