A stochastic action language A

Abstract

In this paper we present a new stochastic nondeterministic high-level action language SAA which is a stochastic extension of Action Language A. We describe the syntax and semantics of SAA and show it has an equivalent expressive power to Hidden Markov Models (HMMs). The main advantage of SAA is its smooth conversion of propositions and probability, and use of a well-established stochastic model. We show two simple examples in the nuclear reactor domain and propose a normalisation technique for declarative probability assignments which match our intuition.