Here is the database structure for a medical MIS table:
CREATE TABLE IF NOT EXISTS `ab_med` (
`med_id` bigint(20) NOT NULL AUTO_INCREMENT,
`med_title` varchar(200) DEFAULT NULL,
`med_posted_date` varchar(200) DEFAULT NULL,
`med_company_name` varchar(200) DEFAULT NULL,
`med_experience` varchar(200) DEFAULT NULL,
`med_salary` varchar(200) DEFAULT NULL,
`med_city` varchar(200) DEFAULT NULL,
`med_description_short` text,
`med_details_link` varchar(500) DEFAULT NULL,
`med_from_city_type` int(1) NOT NULL DEFAULT '1',
`med_from_city` varchar(255) DEFAULT NULL,
`med_collected_time` datetime DEFAULT NULL,
`med_status` int(1) NOT NULL,
PRIMARY KEY (`med_id`),
KEY `MedCity` (`med_city`),
KEY `MedTitle` (`med_title`),
KEY `MedCompany` (`med_company_name`),
KEY `MedExperience` (`med_experience`),
KEY `med_details_link` (`med_details_link`),
FULLTEXT KEY `med_index` (`med_title`,`med_company_name`,`med_description_short`)
) ENGINE=MyISAM DEFAULT CHARSET=latin1 AUTO_INCREMENT=639295 ;
There are around 0.5 million records. A user can search using 2 fields:
Keyword is looked into med_title
, med_company_name
, med_description_short
, med_city
as per the score.
Location is looked into med_city
I've used full text mysql search but I need the most recent entries to show up with better score. I'm not getting the SQL query for it. I'm using the following SQL query:
SELECT distinct(med_details_link),med_id,med_title,med_posted_date,med_company_name,med_description_short,med_from_city,med_experience,med_city,med_collected_time,MATCH (med_title,med_company_name,med_description_short) AGAINST ('+hello +world' IN BOOLEAN MODE) as score
FROM ab_med
WHERE MATCH (med_title,med_company_name,med_description_short) AGAINST ('+hello +world' IN BOOLEAN MODE) order by med_id desc limit 1000
Suppose I search for "hello world", I understand that there should be weight based on:
Please help in getting a SQL query to consider all these weights and search efficiently.
Consider turning to Sphinx to get a satisfactory solution to this problem.
With 5 million rows, and the requirements you have, the performance of MySQL fulltext search will be disappointing.