It has been a while since I posted something. For your comments I add in "bbcsearch".
This system uses a cron based indexer of BBC RSS feeds to a database. The index table is pretty slimline and all common words within a story are removed to reduce database size and improve search times. Currently ive only been indexing for a short period so results would cover more "area" when the database fills.
Front end is powered by Ajax with the data provider producing an XML feed of matched stories. Searches take place two seconds after the last keypress as long as the string is different than the previous.
The ranking algorithm is being worked on and has about 12 different assesments on ranking. Every search has a maximum number of 20 results which are updated within the "ajax app" in real time.
I've tried this with various searches and its pretty accurate, usually coming up with a story im interested in. You should base the search on a target idea such as the recent earthquakes as the predictive nature of the system will produce the results you most want. It would be difficult to judge the system just by typing words without
any preference for results. Please dont hammer it too much, bandwidth maybe limited on this on :) http://www.webcoding.co.uk/bbcsearch Hope you enjoy. James. - Sent via the backstage.bbc.co.uk discussion group. To unsubscribe, please visit http://backstage.bbc.co.uk/archives/2005/01/mailing_list.html. Unofficial list archive: http://www.mail-archive.com/[email protected]/
