Semantic automatization of the data-analytical processes

Abstract

This paper presents the method for the automatization of the data analytical processes using the semantic technologies. The core of the automatization is the machine-readable semantic model, which formalizes the goals of the data analysis, input and output data, possible data operators and data mining algorithms. The proposed semantic model allows automatic composition, orchestration and optimization of the data operators and algorithms in order to achieve specified goals of the data analysis. The evaluation of the semantic model was performed on the two real-world examples where the automatically generated solution was compared with the implementation manually programmed by the data scientist.