SQL Query Recommendation Based on Matrix Factorization

The primary objective of this study is to enhance query processing performance through the integration of sophisticated data reduction techniques within databases. Operating within the domain of in-database analytics, the focus is on evaluating efficient methods for traversing extensive data sets, expediting query processing, and deriving valuable insights using decomposition techniques, specifically QR decomposition. This study aims to achieve two primary goals: firstly, the implementation of decomposition techniques such as QR within a relational database using SQL as an operator, and secondly, the utilisation of column pivoting mechanisms on the reduced data for query recommendation. This amalgamation serves as an experimental pilot space for the innovative hypothesis of integrating query recommendation through the matrix factorization algorithm, QR factorization, using SQL. The core motivation behind the in-database approach for the query recommendation lies in the principle that analysing data where it resides is optimal for the most effective analysis. In this context, query recommendation within in-database analytics streamlines the process of data analysis and exploration by suggesting appropriate queries derived from user behaviour. These recommendations are instrumental in improving efficiency, providing relevant answers, and optimising the overall analysis.