Abstract

Relational DataBase management systems (DBMS) have a great utility in the efficient storage of large data volumes. Also, some DBMS extensions based on fuzzy logic have been proposed to improve the expressiveness of query languages. Among which SQLf is an extension of SQL that supports fuzzy conditions. Separately, the Group-By is a database operator widely used in data analysis and decision support systems. In many cases, it seems useful to group values according to their similarity to a certain concept rather than establishing grouping on the basis of equal values. In this context, a new SQLf structure called Fuzzy Group By (FGB) has been proposed to support a grouping based on fuzzy partitions. In this work, we incorporated the fuzzy grouping in PostgreSQLf, which is an extension of the PostgreSQL DBMS for the handling of fuzzy queries using the SQLf language on the basis of a tight coupled architecture, i.e., directly into the DBMS. We have proposed an algorithm based on a hash to evaluate the FGB operator and also empirically assessed the performance of PostgreSQLf over the TPC Benchmark™ -H (TPC-H).