Causality analysis in factor spaces

The causality from conditional factor to resulted factor is needed to promote the intelligence in factorial data bases.The factor space theory has been presented in last eighties,in which a generalized coordinate system for describing arbitrary objects in the world has been built.According to the framework of factor space,this paper has defined the determining degree and determining region of condition factors with respect to the resulted factor in discrete cases.The reasoning knowledge can be deduced from the determining region successively.An algorithm,named factorial analysis is given in the paper,which mines the causality-relationships in between conditional factors and resulted factor,and forms a causality graph from a piece of factorial analysis table automatically.This method can be widely applied in practice,especially in data mining.Comparing with the existing methods,especially the decision tree,factorial analysis method is easier and more reasonable.