Handling Continuous-Valued Attributes in Incremental First-Order Rules Learning