On the Logic of Probabilistic Dependencies

Judea Pearl

This paper uncovers the axiomatic basis for the probabilistic relation "x is independent of y, given z" and offers it as a formal definition of informational dependency. Given an initial set of such independence relationships, the axioms established permits us to infer new independencies by non-numeric, logical manipulations. Additionally, the paper legitimizes the use of inference networks to represent probabilistic dependencies by establishing a clear correspondence between the two relational structures. Given an arbitrary probabilistic model, P, we demonstrate a construction of a unique edge-minimum graph G such that each time we observe a vertex x separated from y by a subset S of vertices, we can be guaranteed that variables x and y are independent in P, given the values of the variables in S.


This page is copyrighted by AAAI. All rights reserved. Your use of this site constitutes acceptance of all of AAAI's terms and conditions and privacy policy.