This is an idea of AZ. Make sure that longer documents don't skew the results by sampling a fix number of segments from each document. In this way, each document will have an equal weight in the overall Zeta score of a group of texts. It does mean some loss of information, because more text is discarded.
christofs
commented
7 years ago
This is an idea of AZ. Make sure that longer documents don't skew the results by sampling a fix number of segments from each document. In this way, each document will have an equal weight in the overall Zeta score of a group of texts. It does mean some loss of information, because more text is discarded.