Author: Steve
Email: [EMAIL PROTECTED]
Message:
I've been slowly refining the Perl frontend and have found some problems... The most
significant right now is that the current SELECT statement in udmDB module to retrieve
result docs does not handle relevancy correctly. If the following statement is used
instead, things are better:
SELECT url.url, url.title, url.txt, url.content_type, url.docsize,
from_unixtime(url.last_mod_time,'Y-m-d h:i:s') as last_mod_time,
url.keywords, url.description, url.crc32,
dict.url_id,
(SUM(dict.word in ($w)) + SUM(dict.intag)) as r,
($qu) as qu
FROM dict, url
WHERE dict.word in ($w) AND
dict.url_id in ($url_in) AND
dict.url_id = url.rec_id
GROUP BY url_id HAVING qu > 0
ORDER BY r DESC
The important change is that I've added (+) the sum of word weights to the sum of the
number of occurrances.
Any suggestions on further improving this query to better mimic the search executable??
When I complete updating the Perl frontend, I will pass along the revised sources if
that would be desirable to someone.
Reply: <http://search.mnogo.ru/board/message.php?id=2304>
___________________________________________
If you want to unsubscribe send "unsubscribe general"
to [EMAIL PROTECTED]
