Algebraic Bayesian networks minimal join graphcliques comparative analysis

Algebraic Bayesian networks are probabilistic-logic graphic models of knowledge sys-tems with uncertainty and can be applied to statistic data processing and machine learning. Secondary structure usually represented as join graph is crucial for its work. The article represents minimal join graphs cliques classification on the number of their vertexes and the number of special edges they contain. Eight different clique types are designed and estimations of number of components depended on them (feuds and sinews) are obtained and proven.