Automatic Formulation of SQL GROUP BY and HAVING Statements using the Ontology Driven Relational Query Formulation Framework

Abstract

Patterns and relationships within medical data are of great importance for health professionals and clinical researchers. It is a challenging task for clinical researchers to extract interesting patterns from complex medical data, requiring users to develop specific skills in order to create complex database queries. An ontology driven architectural framework, namely OntoQF [1], has recently been proposed to assist users in creating relational database queries. OntoQF works in two phases. In the first phase, OWL ontology is created from relational schema and domain concepts are integrated into it to form domain ontology. In the second phase, domains experts are provided assistance in writing ontology based queries which are translated into equivalent relational queries. The OntoQF was based on OWL1-DL [2] whereas W3C (World Wide Web Consortium) has recently recommended a more expressive and powerful ontology language i.e. OWL2 [3]. OWL2 is an extended version of OWL providing support for more sophisticated ontology constructs and definition of rich data ranges. This work focuses on extending support for some more relational query constructs using selected novel features of OWL2-DL. A set of algorithms is proposed to translate object property cardinality restrictions and data property cardinality restrictions into equivalent relational query restrictions containing GROUP BY and HAVING clauses. Support for novel feature of OWL2 i.e. qualified cardinality restrictions is also provided by proposing an extension to the existing algorithms. In addition, extensions are also proposed to translate quantification restrictions on data properties into equivalent relational query restrictions. Previously, due to the limited expressive power of OWL, support for quantification restrictions was available for object properties only. At the end, all the generated constructs are joined together in a defined manner to create a correct relational query. The generated complex relational queries are validated against a set of test case from medical domain. In conclusion, this work has extended the functionality of OntoQF [1] by providing support for GROUP BY and HAVING constructs using OWL2-DL.