Tweet
I was considering an AJAX search assistant using the OpenSearch RSS from the cgi version, but then it occurred to me that all these potential 2-3 character searches would be dumped in the logs making the search logs rather useless. Have you guys thought about an AJAX search assistant and if so, would you consider designing an optional URL parameter to disable logging for these types of 2-3 character, rapid-fire seaches? I'd wish not to disable search logging for "real" OpenSearch queries.
-
-
I'm not entirely sure how the AJAX search assistant you are describing would work. Are you suggesting that it would send a search query as the user types in one character at a time, and use the results from the RSS/XML feed? I'm not sure if that would do what you want ... wouldn't that just return the result pages, rather than the keywords?
I'm thinking of the search assistant as the "auto complete" sort of feature seen here:
http://www.google.com/webhp?complete=1&hl=en
Is this what you mean?
Because, doing a search per character would be very slow and not suitable for this. The data retrieved would also not be very appropriate.
Perhaps it makes more sense for your search assistant to look in the search logs and get popular searches instead?
You can change the minimum length of words which are indexed on the "Skip Options" tab ("Skip words less than x characters"). But this does not change what is logged, and searches for short words will continue to be logged. -
Definitely sounds like an enhancement request
I don't know if anyone has implemented this yet, but it looks interesting:
http://www.opensearch.org/Specificat...uggestions/1.0
You definitely wouldn't want to perform an entire index search. For an in-house project at my day job, we hit an optimised list of search words / phrases for this kind of AJAX tool.
JasonComment
Comment