I a trying to implement BM25F from scratch in Python. Here is the simplified folumation of the BM25F:
However, after implementing it, I am getting the same score for all the documents in the collection.
After a careful inspection, the only intuition I can understand is that the formulation is considering the stream scores and not the individual document scores.
Which part of the formulation am I missing that considers the individual scores for each document?
I am trying consciously since last week to decode this without any luck, any help would be immensely appreciated!
