Trust-based Enrollment in a Group in a Distributed Setting

Abstract

We investigate how trust can be defined and exploited in the problem of accepting a candidate for enrollment in a group in a distributed setting. A simulation in Python using the NetworkX package has the servers (nodes) implement the WedID protocol. Each node has a profile document with a FOAF graph connecting the server to its neighbors via the foaf:knows property. We assume that the FOAF graphs have been assembled into a network tying together the group. Enrollment relies heavily on two novel notions of trust: one is the trust one node has in another, which we take to come from several kinds of sources, and the other is the trustworthiness a node enjoys because of its centrality. If someone objects to the candidate, there is a vote, a novel procedure that weights a node's votes by its trustworthiness and allows votes based on reference weighted by trust in the referrer. Our main contributions are the procedure for accepting a candidate and the two notions of trust. Results to date include the simulation mentioned; thirty-five test runs provided reasonable results. Work in progress is an implementation of the procedures on a group of servers implementing the WebID protocol.