<![CDATA[Today, the majority of data generated and processed in organizations is unstructured. NoSQL database management systems perform the management of this data. The problem is that these unstructured databases cannot be analyzed by traditional OLAP analytical treatments. The latter are mainly used in structured relational databases. In order to apply OLAP analyses on NoSQL data, the structuring of this data is essential. In this paper, we propose a model for structuring the data of a document-oriented NoSQL database using machine learning (ML). This method is broken down into three steps, first the vectorization of documents, then the learning via different ML algorithms and finally the classification, which guarantees that documents with the same structure will belong to the same collection. Therefore, the modeling of a data warehouse can be carried out in order to create OLAP cubes. Since the models found by learning allow the parallel computation of the classifier, our approach represents an advantage in terms of speed since we will avoid doubly iterative algorithms, which rely on textual comparisons (TC). A comparative study of the performances is carried out in this work in order to detect the most efficient methods to perform this type of classification.]]>
Copyrights © 2025
