Very Computer

Thanks,
Mike

Quote:>Has anyone written a database that has a field containing keywords?  If
>so, what is the best SQL Statement to search on that field?

The trick is a smart algorithm to extract the keywords, and store them
as either the keyword, or some sort of hashcode of it. You may even
store different hashcodes simultaneously, with different degrees of
"compression" applied to them. You'll also need some intelligence to
examine the selection criteria entered and work out the best fields to
compare against; how specific you expect the match to be and whether
to compare against the slightly or highly.compressed hashcodes. If the
user enters a lot of data, be optimisitic and try to get quite a
"tight" match. If they didn't enter much beyond a few characters of a
name, assume the worst and check it against something lax like a
soundex.

When you run the query, use equals. SQL servers, unless you have
something with extensions for free-text handling, just don't do very
well otherwise. If you want good performance, you can do some
processing when you store the original records, you can take your time
building up the query itself, but you must try and avoid operators
such as "Like" which have to do intensive processing on a row-by-row
basis.

--

"Cut the second act and the child's throat"
  - Noel Coward, on seeing the young Bonnie Langford on stage