VanLehn, K. (1989). Efficient specialization of relational concepts. Machine Learning, 4,99-106.

In the process of including a concept from examples, it is sometimes necessary to make a concept less general because the concept matches a negative example. However, one usually wants to reduce the generality as little as possible. Thus, the so-called specialization problem is to find all specializations of a given concept that are maximally general and yet do not match the given negative example. This paper provides an algorithm for a restricted class of specialization problems.

For a pdf of the paper, click here .