On 06/10/11 12:58, François Tissandier wrote:
>> Get them from your web server's log files. Or (better) add some code to
>> your search script to log the search terms separately. We do this to let
>> us add a search cloud pop-up of the top ten most recently-searched terms.
> Mmm good idea, but that's not what I want to do. I want to propose the
> most popular keywords from my content, not from the searches !
Ah. But how are you defining "most popular" if not by what is searched for?
> And those
> keywords are in the index, so I thought there is maybe a way to extract
> the most popular ones. By "popular" I mean "keywords appearing the most
> often". Sorry if my frenglish is not clear !
OK. These are not "keywords" at the moment: they are just words.
Keywords are words that you have preselected for some special reason, so
they will become keywords once you have identified them.
You need to crawl the site (or from the server, just traverse the
document root), strip any markup, split into one word per line, do a
count by word, eliminate stopwords, and rank the rest.
Users mailing list
Received on Thu Oct 20 2011 - 12:10:09 GMT