April 18
Vitor Santos Costa
COPPE/Sistemas, Universidade Federal do Rio de Janeiro
CLP(BN): Constraint Logic Programming for Probabilistic Knowledge

Recent work by Koller, Getoor and others on Probabilistic Relational
Models (PRMs) has shown the advantages of approaches based on
relational logic to describe statistical models of structured
data. CLP(BN) is a novel logic programming system, inspired on the
PRMs, where uncertainty over the attributes of an object is
represented as constraints. CLP(BN) permits recursion, the use of
functor symbols, and the representation is amenable to learning using
techniques from inductive logic programming.