Markov Logic Networks are a popular probabilistic-logic modeling framework based on un-directed Markov Networks. It has been shown in this paper how Markov Logic Networks can be compiled into equivalent Relational Bayesian Networks. Primula now contains an implementation of this translation. For example, a MLN model like friends_and_smokers.mln is parsed by Primula, and compiled into its RBN equivalent friends_and_smokers.rbn , which then enables all Primula's exact and approximate inference techniques to be applied to this model:
In particular, Primula's exact inference techniques complement (for small domains) the Alchemy System's more sophisticated approximate inference techniques.