I wuld like to do som automatic tagging of incoming text in our system and I was wondering if a full-text index is capable of providing a a ranked list of words given an indexed row.
If not, do you have any suggestions on how to do this? We already have a system in place for auto tagging but its fairly brute-force (aka. slow) in its method and it is not always yielding a good list of words.
I think so. From here
The scenarios given for using sys.dm_fts_index_keywords_by_document are
“I want to know how many keywords the full-text index contains”
“I want to know if a keyword is part of a given doc/row”
“I want to know how many times a keyword appears in the whole full-text index” (sum(occurrence_Count) where keyword=……)
“I want to know how many times a keyword appears in a given doc/row”
“I want to know how many keywords a given doc/row contains”
“I want to retrieve all the keywords belonging to a given doc/row”