2.1 The main cycle of the program

YAILS consists of an examples processing cycle which learns a theory starting with a possibly empty one. This means that at the end of each iteration we have a theory that "explains" the seen examples. This can be described by the following algorithm :

FOR each example DO

Try to incorporate it into the present theory (Modification of the theory)

Calculate the completeness of the modified theory

IF it is less than demanded by the user THEN

Invent new rules until it reaches the desired value

One of the important notions presented by the algorithm is the completeness of a theory. In this report I consider the completeness of a theory the ratio between the number of examples covered by all rules of the theory divided by the total number of examples seen by the algorithm (see appendix A for a summary of several important notions that are used throughout this document). Notice that the user can control the level of completeness.