Discovery of domain principles has been a major long-term goal for scientis
ts. This paper presents a new system named DOMRUL for learning such princip
les in the form of rules. A distinctive feature of the system is the integr
ation of the certainty factor (CF) model and a neural network. These two el
ements complement each other. The CF model offers the neural network better
semantics and generalization advantage, and the neural network overcomes p
ossible limitations such as inaccuracies and overcounting of evidence assoc
iated with certainty factors. It is a major contribution of this paper to s
how mathematically the quantizability nature of the CFNet since previously
the quantizability of the CF model was demonstrated only empirically. The r
ule discovery system can be applied to any domain without restriction on bo
th the rule number and rule size, In a hypothetical domain, DOMRUL discover
ed complex domain rules at a considerably higher accuracy than a commonly u
sed rule-learning program C4.5 in both normal and noisy conditions, The sca
lability in a large domain is also shown. On a real data set concerning pro
moters prediction in molecular biology, DOMRUL learned rules with more comp
lete semantics than C4.5.