Update app.py

app.py

@@ -29,11 +29,12 @@ def read_pdf(file):
   # We use BM25 as retriver which will do 1st round of candidate filtering based on word based matching
 
 def bm25_tokenizer(text):
+    stop_w = ['a', 'the', 'am', 'is' , 'are', 'who', 'how', 'where', 'when', 'why']
     tokenized_doc = []
     for token in text.lower().split():
         token = token.strip(string.punctuation)
 
-        if len(token) > 0 and token not in _stop_words.ENGLISH_STOP_WORDS:
+        if len(token) > 0 and token not in stop_w:
             tokenized_doc.append(token)
     return tokenized_doc