exact match scoring higher

[Ronald Christian]

In the search function of a knowedge base, I'm doing a full text match in binary mode, ordered by score, descending.

If I do a query "+linux +user +management", and eyeball the results, it looks like the results are ranked by the number of occurances of the words "linux", "user", and "management".

If I do a query "linux user management", I get only the two or three articles that have that literal string.

What I'd like to see is all the articles that have all three words, but ranked such that the exact match is rated highest. In other words, we want to see the rest of the results, but it's the exact match that is the most interesting.

My solution is to do two queries, displaying the results of the "linux user management" query followed by the results of the "+linux +user +management" query. This is inefficient and gives duplicate results. I'm hoping that the experts here can show me a better solution.