Processing of data for inductive inference based on non-strict probability

Based on methods of inductive logic, an approach to identifying of implication relationships “If A, then b” in Big Data is considered. This approach is considered in conditions of low reliability and inconsistency of data. To work in this condition, logics with vector semantics in the form of VTF logics are used. The presence or absence of phenomena in tables of their joint occurrence is formalized by truth vectors with components v$^+$ and v$^-$, where v$^+$ is a measure of the true of a statement about the presence of a phenomenon, v$^-$ is a measure of its false. On the base of statistical induction principal, the indicator of the validity of a causal relationship is calculated as the average value of the truth vectors of the corresponding non-strict propositions. The resulting value is interpreted as a non-strict probability of the relationship, which acts as a vector indicator of its validity. The applicability of the approach for processing qualitative and quantitative data, as well as data containing artifacts, is shown.