When introducing new document, fast calculation of ft-idf is needed. The issue is the re-scanning of all old documents for the new tokens for idf calculation. So, if the token h
