Copy the page URI to the clipboard
Grimsley, Micheal and Meehan, Anthony
(2002).
DOI: https://doi.org/10.1007/3-540-45741-0_11
URL: http://www.metapress.com.libezproxy.open.ac.uk/con...
Abstract
We describe a metric to assess agent trustworthiness from the earliest stages of a dialogue between two web agents. There is no assumption of a transaction history between the agents nor is there a requirement for the agents to fully share the semantics of the set of alternatives over which negotiation occurs. The metric is designed to recognise a form of co-operative negotiation behaviour, so-called logrolling, which is known to induce trust between human negotiators. The metric requires an agent to be able to infer the issue priorities of the other party over a series of proposals and to correlate these with its own priorities. An example is used to illustrate how this may be achieved.