drop duplicates when showing documents (#6)

gradio_app.py

@@ -34,6 +34,10 @@ def format_sources(matched_documents: pd.DataFrame) -> str:
     matched_documents.similarity_to_answer = (
         matched_documents.similarity_to_answer * 100
     )
+
+    # drop duplicates, keep highest ranking ones
+    matched_documents = matched_documents.sort_values("similarity_to_answer", ascending=False).drop_duplicates("title", keep="first")
+
     documents = "\n".join(
         [
             document_template.format(document=document)