Probabilistic, Logical and Relational Learning - A Further Synthesis