With the growing prevalence of XMPP communications, it is no longer safe to assume that XMPP will be free of spammers and nuisance users due to its obscurity. To that effect, this protocols aim is to create a user rating that can be used to hinder unnecessary communication within servers. While there have been several attempts to curb spam within servers, they have targeted the specific messages that may be labeled as spam, or relied on blocking policies. This approach singles out the JID of the user to (hopefully) remove their access to the server.

XMPP Core, not known at this point.

A user rating is an integer value assigned to a user’s bare JID to aid servers in determining JIDs that are predisposed to spam or offensive messages. The User Rating allows servers to determine appropriate actions from limiting stanza sending, to kicking or banning of the JID from the server. The rating, once implemented can be used in automated and user-based environments to curb abusive users. element should be attached to a bare JID and should contain a float. 0.0]]> Would indicate the user is behaving normally, and is not suspected of sending spam. Where a user who has achieved a rating of the following: 1.0 would be kicked from the server, for example. is the JID of the user to be reported for spam or abuse. ]]>

OPTIONAL.

element may be incorporated into non-human based systems as well. Unexpected JIDs, or ones that connect externally may be automatically given a high rating, and the server set for a low tolerance of stanzas sent in, allowing for quick and generally automated expulsion of unwanted or abusive JIDs from an automated system. The element does not require a human interaction, instead it can be manipulated by a host server for automated abuse control. Depending on a server operator’s level of forgiveness, ratings may be permanent or can normalize to 0 after a period of time.]]>

This feature may lend itself to abuse from users who wish to punish or abuse another user. All rating systems should follow two abuse control rules to limit abuse potential:

1. No one entity can use the rating system to ban/kick a JID unless they are a server administrator. There must at least two or more JIDs reporting spam or abuse to pass the kick or ban threshold.
2. Successive report values on same JID will decrease in weight against a user rating, eventually lowering to zero (which then can be marked as spam/abuse). For example: the first <repored-jid/> of mercutio@montague.net by Romeo@montague.net increases the rating by 0.1. The second from romeo@montague.net the same user is now weighted slightly less, by increasing it 0.08. The third 0.06, fourth 0.04 and so on until the weighted value of the report is null. At this point, it would be advisable to notify Romeo that he is abusing the rating system, and will begin to increase his own rating by further pushing the issue.
3. Critical or known-user JIDs exempt from rating rules should be set to a value such as -100 to indicate that they are exempt from spam control measures.

This document requires no interaction with the Internet Assigned Numbers Authority (IANA).

None known at this point.

Not finished at this point.