Remove plural names

entity_extraction.py

@@ -3,6 +3,15 @@ import spacy
 nlp = spacy.load("en_core_web_md")
 nlp.add_pipe("entityfishing")
 
+def remove_plural_names(article):
+    words = article.split()
+    new_words = []
+    for word in words:
+        word = word.replace("’s", "")
+        word = word.replace("'s", "")
+        new_words.append(word)
+    return " ".join(new_words)
+
 
 def extract_entities(article):
     '''Find wikidata refs for article entities'''
@@ -11,6 +20,7 @@ def extract_entities(article):
     seen_surnames = []
     seen_qids = []
 
+    article = remove_plural_names(article)
     doc = nlp(article)
     for ent in doc.ents:
         if ent._.kb_qid is None or ent.label_ not in ["ORG", "PERSON", "GPE"] or ent.text in seen_entities: