I've revised the tf() function to always return 1, regardless of the number
of terms it finds.
However, I run into a problem when a stemming words and root words appear
together. These documents get a higher boost than documents with just the
root.
For example: "woman walking fast" gets tf(woman) = 1
"woman walking fast women walking fast" gets tf(woman) = 1
and tf(women) = 1, resulting in higher score than just "woman"
Is there a way to always return 1 for tf(), regardless of stemming words or
synonyms?
Thanks,
Hung
