Logic, Probability and Computation:Foundations and Issues of Statistical Relational AI

摘要

Over the last 25 years there has been considerable body of research into combinations of predicate logic and probability forming what has become known as (perhaps misleadingly) statistical relational artificial intelligence (StaR-AI). I overview the foundations of the area, give some research problems, proposed solutions, outstanding issues, and clear up some misconceptions that have arisen. I discuss representations, semantics, inference and learning, and provide some references to the literature. This is intended to be an overview of foundations, not a survey of research results.