Fuzzy query translation from sql to Mongo QL

Abstract

Relational DBMS are often used to store fuzzy values, but there are problems with placing such data in a tabular form. Besides, there appears the problem of storing both the clear and fuzzy data related to one subject area in one relational table column. This article considers the mechanism of storing clear and fuzzy values and linguistic variables in the documented Mongo DBMS. The data are stored in the collection as GeoJSON geometry, different geometries are used for different data variants. The possibility of storing clear scalar values, clear value sets, clear value intervals and fuzzy values in the collection documents is described. For data processing by means of SQL queries, the context-free grammar of the SQL subset is described, which generates lexer and parser. To form the structure of the abstract syntactic tree the corresponding object model is implemented. A translator application has been developed which allows converting SQL queries based on clear and fuzzy data into Mongo QL queries. The algorithm of fuzzy queries translation process is offered, the geometrical interpretation of data comparison operations is described. The examples show the variants of fuzzy comparison operations for different variants of values.