Contrôle de qualité de données NoSQL

Abstract

This work revolves around quality control of NoSQL data, specifically document-oriented data. In fact, it relies on a method that allows to detect and repair the problems of schematic overlap, duplication, and incompleteness based on the frequency of database elements. For this purpose, a new method named MFU (Most Frequently Used) has been proposed. The MFU method consists of three phases: (1) quality problem detection, (2) data repair, and (3) quality verification. In each phase, three types of data quality problems are addressed. Our MFU method has been validated by implementing the Quality of Document oriented Database (QoDB) tool and evaluated on the MongoDB COVID19 database that was released in 2022. The results obtained are interesting.