Fuzzy SQL queries in standard SQL database

Abstract

Uncertain queries are very common in many areas of human activity. The problem can be seen particularly in medicine, where expressions like ”very high”, ”low”, ”normal” are commonly used in order to describe different information. However, the most popular data repositories do not allow to form imprecise queries in order to filter information. Therefore, the paper proposes an extension to the standard SQL language allowing anybody to profit from fuzzy database using any SQL engine. The existing approaches employ different mechanisms in order to allow the user to perform fuzzy queries on a database. The most complex solutions modify the database engine itself. However, such approach is strongly bound to the modified server version and must be updated with any development of the original server. Nevertheless, there is possible to store fuzzy information using for instance columns of regular relational database. Therefore, this approach proposes extensions to the query language allowing to use fuzzy information in a query and provides a parser transforming a fuzzy query into a standard SQL. Thus, the database server version is irrelevant. The solution is provided as a module written in multi-platform Java language using popular JDBC database connection.