Evolving SQL Queries from Examples with Developmental Genetic Programming

Abstract

Large databases are becoming ever more ubiquitous, asare the opportunities for discovering useful knowledgewithin them. Evolutionary computation methods such asgenetic programming have previously been applied toseveral aspects of the problem of discovering knowledgein databases. The more specific task of producinghuman-comprehensible SQL queries has several potentialapplications but has thus far been explored only to alimited extent. In this chapter we show howdevelopmental genetic programming can automaticallygenerate SQL queries from sets of positive and negativeexamples. We show that a developmental geneticprogramming system can produce queries that arereasonably accurate while excelling in humancomprehensibility relative to the well-known C5.0decision tree generation system.