[foaf-protocols] Structured Secure Messaging over FOAF

Melvin Carvalho melvincarvalho at gmail.com
Mon Sep 14 16:11:44 CEST 2009


I've been thinking a bit more about structured messaging over FOAF.
This is the model that I think could work.


Underlying FOAF
==============

FOAF could be structured in 3 "layers"

1. Raw Data (RDF)
2. WOT Data
3. Access Control Data

In this way you have a visible set of structured data that is
available for other agents to view, and to allows you to endorse
certain statements and also to provide permissions.

Each layer is built on top of the other (I put WOT at level 2, because
i think you may want to access control that too, and also this will
become important for the social web to have trust and self healing)


Authentication
============

This can be done securely with SSL/Cookie/Shared Secrets, either
client server, or server to server


Messaging
=========

I see this as being the value add part of the system, once you've
built out the secure structured components above.  I would suggest
SPARUL as a messaging system as this will allow the richness creating
a set of triples to have virtually any meaning.

Some care must be taken in the ordering of messages because they are
likely to be asynchronous and both agents may not always be online.
Therefore, a queuing mechanism will normally be built in.
Additionally an approvals and notifications system will normally be a
policy to present users with an approve/reject button in certain
instances (a good example of an implementation this is adding a friend
to your roster in IM systems)


Some of this stuff will need more fleshing out, but I think that this
model could form a pretty good infrastructure for secure structured
messaging (and trust) on the web.

Did I miss anything?


More information about the foaf-protocols mailing list