Implementation of a Transformation System for Relational Probabilistic Knowledge Bases Simplifying the Maximum Entropy Model Computation.
Christoph Beierle, Markus Höhnerbach, Marcus Marto · The Florida AI Research Society · 2014
The maximum entropy (ME) model of a knowledge base R consisting of relational probabilistic conditionals can be defined referring to the set of all ground instances of the conditionals. The logic FO-PCL employs the notion of parametric uniformity for avoiding the full grounding of R. We present an implementation of a rule system transforming R into a knowledge base that is parametrically uniform and has the same ME model, simplifying the ME model computation. The implementation provides different execution and evaluation modes, including the generation of all possible solutions.