Semantically Refining the Groundwater Markup Language (GWML2) with the Help of a Reference Ontology

Abstract

GIScience 2016 Short Paper Proceedings Semantically Refining the Groundwater Markup Language (GWML2) with the Help of a Reference Ontology Torsten Hahmann 1 , Shirly Stephen 1 , Boyan Brodaric 2 School of Computing and Information Science, University of Maine, Orono, ME 04469, USA Email: {torsten.hahmann | shirly.stephen}@maine.edu Geological Survey of Canada, Ottawa, Canada K1A0E9 Email: boyan.brodaric@canada.ca Abstract Reference ontologies are intended to aid domain ontology design, identify gaps and inconsist- encies in representations of domain information, and facilitate data interoperability. The appli- cation of a reference ontology to the water domain is untested. We present findings from using a first-order logic reference ontology for the water domain, the Hydro Foundational Ontology (HyFO), to identify and remedy semantic gaps and inconsistencies in the Groundwater Markup Language (GWML2), a data model for groundwater information with less detailed formal se- mantics. We express GWML2 as a logical extension of HyFO, thereby improving GWML2’s compatibility with other hydro data models. We derive general desiderata for a “good” domain reference ontology in the geosciences and discuss the benefits one can expect from their use for the ontological analysis of geoscience data models. 1. Introduction Effective water management requires exchanging and integrating information about the lo- cation, quantity, and flow of water throughout the water cycle. The information is typically stored in multiple data stores based on different data structures, terminologies, and light-weight ontologies, subsequently summarily referred to as data models. Knowledge integration and querying across these data stores requires interoperability between their representations at the syntactic, schematic, and semantic (comprising differences in terminology and definitions) lev- els. To prepare for automated integration of geoscience knowledge across these levels, we ex- plore the use of a reference ontology (Noy 2004) as a tool for increasing semantic precision and coherence in geoscience data models. We specifically test this idea within the hydro domain by using the Hydro Foundational Ontology (HyFO), a reference ontology for the hydro domain developed since 2011 (see e.g., Hahmann & Brodaric 2012, 2013; Brodaric & Hahmann 2014), to semantically analyze the Groundwater Markup Language (GWML2) (Boisvert & Brodaric 2012; Brodaric 2015) as one example of a hydro data model developed by domain scientists. The result is an improved version of GWML2 with (1) increased semantic precision through axiomatic constraints (i.e., constraints expressed in a logical language such as first-order logic) and definitions based on well-defined reference terms from HyFO, (2) a stratified formalization that separates concepts based on how broadly they apply (across geosciences, to the entire water domain, or only to groundwater), and (3) a completed taxonomy that fills gaps and renames classes to better reflect their position within the stratification. The domain reference ontology HyFO is not yet another standard that restricts fairly generic scientific terms (classes and relations), such as geologic unit, water body, or aquifer, to a single interpretation. Instead, it provides a neutral but concise language for describing in a machine- interpretable format how terms are used in existing data models. Such formal descriptions sub- sequently allow data models to be semantically compared and integrated in a largely automated fashion for integrated querying and knowledge discovery as envisioned in semantic e-science (Brodaric and Gahegan 2010).