Algorithms for rule inference in modularized rule bases

Abstract

In the paper an extended knowledge representation for rules is considered. It is called Extended Tabular Trees (XTT2) and it provides a network of decision units grouping rules working in the same context. The units are linked into an inference network, where a number of inference options are considered. The original contribution of the paper is the proposal and formalization of several different inference algorithms working on the same rule base. Such an approach allows for a more flexible rule design and deployment, since the same knowledge base may be used in different ways, depending on the application. © Springer-Verlag Berlin Heidelberg 2011.