Modeling Dependent Events with CHRiSM for Probabilistic Abduction
Henning Christiansen, Amr Hany Saleh · 2011
Most earlier approaches to probabilistic abductive logic pro- gramming are based on the assumption that abducibles represent in- dependent events. This enables ecient and incremental calculation of probabilities, but may not be suitable for all real world problems. As an attempt to introduce such dependencies in a logic programming setting, we have applied CHRiSM, which is a recent probabilistic extension to Constraint Handling Rules, for specication and evaluation of probability distributions over dependent abducibles. It is shown that this principle integrates well with earlier work on probabilistic abduction based on CHR, generalizing it to handle such dependencies. We present our rst