in reply to Fulltext DB search: The Need for Speed
One heuristic is to create an internal stoplist of common words and look up only non-stoplisted words first. Then from that small set of matching records, search for the stoplisted words. Set intersection is associative -- use that to your advantage.
Another heuristic is to use caching. If there are a relatively small number of very common and commonly requested words, create a table of matching records for each word and search only that table if the word is in the phrase. This prevents having to pull up most of a table each time such a word appears.
-Mark
|
|---|