This is a strong argument for a client-side tagging component (e.g. a Firefox extension). There are other advantages to this as well, such as the fact that you can actually see the page you are tagging. You can do a lot of neat stuff when you have the entire DOM for the page parsed and in memory.
I've been meaning to add support for a pages META keyword tags to Scrumptious. Maybe I'll fiddle with a bit of automatic tag extraction as well. If anyone knows existing research and/or open-source code relating to this, please let me know. Cheers, Matt > -----Original Message----- > From: [EMAIL PROTECTED] [mailto:[EMAIL PROTECTED] On > Behalf Of Sam Joseph > Sent: Friday, June 03, 2005 2:31 AM > To: discuss@del.icio.us > Subject: Re: [delicious-discuss] Recommended tags based on a site's URL > > In the NeuroGrid system I used to extract all the words from a page, > remove stopwords and then present the most frequently occuring terms to > the user as tag possibilities. More sophisticated approaches might use > TFIDF or something like that. > > The main problem with this, and indeed any other approach that involves > parsing the page in question, is the time it takes to get the page and > parse it. The parsing usually won't take so long, but there will be a > few seconds delay to grab the page, and this can be a little frustrating > to users who are expecting a quick turnaround such as they currently get > with del > > CHEERS> SAM > > Clifford Caoile wrote: > > >But, how would you propose getting the recommendations in the first > >place? What algorithms can succinctly summarize a home page? > > > > > > > _______________________________________________ > discuss mailing list > discuss@del.icio.us > http://lists.del.icio.us/cgi-bin/mailman/listinfo/discuss _______________________________________________ discuss mailing list discuss@del.icio.us http://lists.del.icio.us/cgi-bin/mailman/listinfo/discuss
