Program synthesis from Workflow-Driven Ontologies

Abstract

An approach that results in the development of Workflow-Driven Ontologies (WDO) (called the WDO approach) allows domain scientists to capture process knowledge in the form of concepts as well as relations between concepts. Program synthesis techniques can be employed to generate algorithmic solutions by transforming the process knowledge expressed in the WDO as concepts and relations to variables and functions and computing unknown variables from known ones based on the process knowledge documented by the domain scientist. Furthermore, the algorithmic solutions that are generated by program synthesis potentially can support the composition of services, which result in the creation of executable scientific workflows. The ultimate goal of this work is to provide an end-to-end solution for scientists beginning with modeling the processes for creating work products in terminology from the scientist's own domains and ending with component-based applications that can be used to automate processes that can advance their scientific endeavors. These applications can exploit distributed components that are supported by current cyber-infrastructure efforts. This paper discusses extensions to the WDO approach that support program synthesis. To elucidate this scenario, an example related to earth sciences is presented.