Howdy,
I recently rolled a custom WordNet synonym filter that pulls synonyms
from WordNet during indexing. All that is nice and dandy; however, it
causes problems in the sorting. Sometimes, the top match will come from
a synonym rather than the original word.
An example in our system is a search on the word "party". The first
match was a doc with the word "company" which, in WordNet, has "party"
as a synonym.
So the question is: is there a way to weight original word matches more
than synonym matches? In other words, I want all docs with "party" in
the text to come up first, then docs whose terms have "party" as a synonym.
I'm hoping there is a way ;) It would seem this would be a common enough
issue. However, from my looking at the source code I fear the worst as I
don't see that you can do this unless it would make sense to utilize the
"type" field of Token.
Thanks,
Lajos
