Composition of Biasses of Inductive Logic Programming

. Studying biases is an important topic in Inductive Logic Programming,and many works have dealt with the problem of dening interestingformalisms for expressing biases. The expected properties forsuch representation languages are usually declarativity, universality, easyshift of biases and learning optimization. In this paper, we propose a newformalism which was designed when attempting to apply learning techniquesto Data Mining applications. In that context, an important property...