PDL2 handles many of the aspects of the
brittleness problem, as characterized in [20]. Its mode
of execution, that combines rule-matching and
similarity-based reasoning allows it to deal with a lack
of a matching rule. When confronted with a new
situation, the system is never left in a "don't know"
state. A conclusion is always reached, either by
applying a matched rule, or by using similarity with
other learned situations.