Update app.py

app.py

@@ -20,17 +20,34 @@ model = pipeline('token-classification',model='neural-net-rahul/bert-finetuned-n
 
 def predict(text):
   result = []
+  word1 = None
+  entity_past = None
   for dicti in model(text):
     entity,word = dicti['entity'],dicti['word']
-    if entity == "B-PER" or entity=='I-PER':
-      entity = "Person"
-    elif entity == "B-LOC" or entity=='I-LOC':
-      entity = "Location"
-    elif entity == "B-ORG" or entity=='I-ORG':
-      entity = "Organization"
-    elif entity == "B-MISC" or entity=='I-MISC':
-      entity = "Miscellaneous"
-    result.append({entity,word})
+    if entity[0]=='B':
+      if word1 is not None:
+        if entity_past =='B-PER':
+          entity_past = 'Person'
+        elif entity_past =='B-ORG':
+          entity_past = 'Organization'
+        elif entity_past =='B-MISC':
+          entity_past = 'Miscellaneous'
+        elif entity_past =='B-LOC':
+          entity_past = 'Location'
+        result.append([word1,entity_past])
+      word1 = word;
+      entity_past = entity;
+    else:
+      word1 = word1 + word.lstrip("#");
+  if entity_past =='B-PER':
+    entity_past = 'Person'
+  elif entity_past =='B-ORG':
+    entity_past = 'Organization'
+  elif entity_past =='B-MISC':
+    entity_past = 'Miscellaneous'
+  elif entity_past =='B-LOC':
+    entity_past = 'Location'
+  result.append([word1,entity_past])
   return result
 
 gr.Interface(