3.4 "Normal", "only foreground" and "all" strategies

YAILS has three different classification strategies that can be used to classify a set of examples. The first is the "normal" classification procedure which is the standard classification algorithm used by YAILS. This procedure involves the division of the learned theory into two sub-sets, the previously referred FS and BS. After this division the classification is obtained using the following algorithm :

Search in FS for an answer taking into account the Minimal Matching Score IF no answer was obtained THEN

Search in BS for an answer taking into account the Minimal Matching Score

IF no answer was obtained THEN

Search in FS for an answer with Minimal Matching Score equal to 0

IF no answer was obtained THEN

Search in BS for an answer with Minimal Matching Score equal to 0

With respect to the second strategy ("only foreground") it consists on using only the initial foreground set. This means that after the division YAILS never looks into the BS to classify the examples.

Finally the last classification procedure uses all the learned rules (maximum redundancy) without doing any division.

These different possibilities enable the observation of the importance of redundancy in a particular problem [Torgo,1992].