1. build an indexer which takes all the words used in the search and put them into a table with these columns.
word, record_id, word_count, word_in_title
word == the word itself
record_id == where the word points to
word_count == frequency of word in document
word_in_title == does word exist in title
searching for terms than goes like this( notice that I only put the % at the end of the like so it will still use an index! )
select record_id, sum(word_count), max(word_in_title) from TABLE where word like '$word%' group by record_id
if the user searches for multiple words than run the above statement multiple times and sum them up by record_id.
Spit them back out by the max word_count.
My example site: www.historychannel.com ( does no caching of results, still very quick )
<code> <a> <b> <big> <blockquote> <br /> <dd> <dl> <dt> <em> <font> <h1> <h2> <h3> <h4> <h5> <h6> <hr /> <i> <li> <nbsp> <ol> <p> <small> <strike> <strong> <sub> <sup> <table> <td> <th> <tr> <tt> <u> <ul>