Probabilistic dispatch, dynamic domain architecture, and self-adaptive software

Abstract

In this paper we report on a beginning effort in the self adaptive software research area of improving function or method dispatch. We extend type-signature based method dispatch in a dynamic object oriented programming language with probabilistic dispatch, where the choice of method to use is determined by statistical means. This research direction is part of a larger self adaptive software effort at the MIT Artificial Intelligence Laboratory, called Dynamic Domain Architectures. © Springer-Verlag Berlin Heidelberg 2003.