Algebraic Bayesian Networks: Probabilistic-Logic Inference Algorithms and Storage Structures

Abstract

This article presents math library and relational database, being components of software complex, that implements latest theoretical results in the field of Algebraic Bayesian Networks storage structures and probabilistic-logic inference algorithms. A review of existing software implementations of given algorithms is performed and a number of deficiencies that are present in them is identified. The description of probabilistic-logic inference tasks and the corresponding public methods of the mathematical library is proposed in the article. The structure of the math library is represented by a class diagram and supplemented by a propositional formula parser algorithm description. Acomparison of existing database solutions based on their domain area is conducted and the substantiation of the choice is provided. The structure of the database is described in detail and a parallel between the theoretical objects and databasetables is drawn. Moreover problems that have arised during the development are considered and an appropriate solution is provided.

Keywords

• algebraic Bayesian networks
• probabilistic-logic inference
• alternative models of knowledge patterns
• matrix-vector equations
• knowledge with uncertainty
• relational databases