Does the use of Data Mining lead to other results than with a normal Case-Based Reasoner using a k -Nearest Neighbor algorithm to compute similarity metrics?