Natural Language Processing in Translation of Relational Languages

Abstract

Methods of data manipulation in combination with sturcture of the data and integrity constraints define data model used in the relational databases. This article focuses on the methods and processes of operation sets which are used for selection of data from relational database and translation between various formats of this manipulation. The article presents the design, implementation and experimental evaluation of tool for translating between relational algebra, tuple relational calculus, Structured Query Language and unrestricted natural language in all directions. Presented software tool translates from natural language with the use of artificial intelligence methods for natural language processing, syntactic analysis and parsing. Generative pretrained transformer of third generation (GPT3) based Codex – Davinci translation model was used as an artificial intelligence method used in the workflow of translation from English language to relational languages.