Mistakenly not sent to the list...
-------- Forwarded Message -------- Subject: Re: [Standards] Reputation system for XMPP Date: Mon, 4 Sep 2017 20:17:31 -0600 From: Peter Saint-Andre <[email protected]> To: Evgeny Khramtsov <[email protected]> On 9/4/17 1:33 AM, Evgeny Khramtsov wrote: > Sun, 3 Sep 2017 17:07:32 -0600 > Peter Saint-Andre <[email protected]> wrote: > >> I did some reading about reputation systems last year, and my >> recollection is that they are not proven to help all that much with >> spam compared to the cost of implementing them, but I'm very much >> open to better evidence. > > Sorry, I misread this. > I somewhat disagree. We see Google's CAPTCHA is using some information > about the entity being challenged (i.e. about its "past behavior"), > which is somehow rated, therefore it's clearly a reputation system. > I myself don't have any evidence though, the only evidence provided are > simulations from those research articles I pointed earlier. In this > sense, of course, reputation systems efficiency is not proven. A friend of mine at PayPal once described to me a bit about all the information that goes into a trust algorithm they use to prevent account hijacking (logging in from an unusual location and so on). We could see that as a kind of reputation, I suppose. The information I was reading can be found here: https://wiki.asrg.sp.am/wiki/Main_Page There is a lot of research on these topics, such as: http://smg.media.mit.edu/library/resnickReputationACM.pdf https://tools.ietf.org/html/rfc7070 https://en.wikipedia.org/wiki/Anti-spam_techniques https://en.wikipedia.org/wiki/Reputation_system The mailing list archives of the ASRG also make for interesting reading: https://www.ietf.org/mail-archive/web/asrg/current/maillist.html See for instance the thread starting here: https://www.ietf.org/mail-archive/web/asrg/current/msg16604.html Peter
signature.asc
Description: OpenPGP digital signature
_______________________________________________ Standards mailing list Info: https://mail.jabber.org/mailman/listinfo/standards Unsubscribe: [email protected] _______________________________________________
