[freenet-devl] freegle.com: major overhaul -> NeuroGrid

[Sam Joseph]

Stefan Wrote: >From: "Sam Joseph" <[EMAIL PROTECTED]> > > I'm not clear about how Freegle works out which keywords to associate > > with the keys (looked in the FAQ and about Freegle, but no joy), but the > >scheme I describe above is implemented in aggregate by DirectHit.com and > > in a personalized manner by NeuroGrid.com (my own offering). The > > NeuroGrid keyword-file association tracking mechanism is already being > > implemented in a couple of other P2P systems. > Currently Freegle matches search keywords with the key itself. In case of > HTML pages, the title is also stored and displayed. Making it searchable too > is not very difficult. The same can (and will) be done for meta > keywords/description and even the content of the page itself. When you say that Freegle matches search keywords with the key itself, you mean that only words that appear in the key are used? So if I search for "Douglas" then only those keys that contain the string "Douglas" will be returned, plus those HTML pages that have titles specified? > Of course, this doesn't prevent malicious publishers from using > inappropriate keywords - as it happens on the WWW all the time. Sure > Using feedback for rating is close to impossible with a Freenet search > engine the way it is currently organized. The web variant could track clicks > on search results via a simple redirect script. But the in-Freenet variant > can't track anything. Really? Isn't the key index used by Freenet search engines also saved into Freenet? What does that file consist of? Just the keys themselves, or keys followed by a list of associated keywords? If the key index is being updated with new keys and being restored in Freenet couldn't you be updating degrees of association of keywords with keys based on user feedback? e.g. utdqti3tro7qiet6atsql2iugakfyg Douglas (6), Adams (5), Hitch-Hikers(4) I mean effectively you would be storing what NeuroGrid stores efficiently in a relational database in a flat file, but you could still do it, no? > Also, in the case of deliberately misleading pages, users don't know they've > been fooled before they actually load the page (and then it's too late for > the search engine to notice). Which is why NeuroGrid is set up, or will be set up, to record bookmarking activity after things have been downloaded, for HTML pages, if you didn't bookmark it after you searched for it, then you probably weren't that pleased with what you found. This is a general problem in terms of getting feedback on search results, for things other than bookmarks I think people might will use a big red "this is not what I wanted" to jump back to the search list when they get something undesirable. On the other hand, if a NeuroGrid Freegle system was set up to associate additional keywords with keys then people would be able to tell quickly if there search for "Marx" was only generating hardcoreporn.mpeg4 results, and not click through on the keys that didn't seem to match their search. CHEERS> SAM http://www.neurogrid.com _______________________________________________ Devl mailing list [EMAIL PROTECTED] http://lists.freenetproject.org/mailman/listinfo/devl