Abstracting Transport to an Ontology Design Pattern for the Geosciences Brandon Whitehead1 , Benjamin Adams1 , Mark Schildhauer2 , Charles Vardeman3 , Werner Kuhn4 , Adam Shepard5 , and Krishna Sinha6 1 Centre for eResearch, University of Auckland 2 National Center for Ecological Analysis and Synthesis, UCSB 3 Center for Research Computing, University of Notre Dame 4 University of Münster 5 Woods Hole Oceanographic Institution 6 Virginia Polytechnic Institute and State University Abstract. A core concept in geoscience/physical science research is the concept of transport. We present an ontology design pattern for the no- tion of transport in the geosciences using natural language coupled with a concept map. The top level concepts of the Transport Pattern are trans- port entity, transport mechanism, and transport event. These concepts are described in detail, and a brief example is provided to illustrate the usefulness of the pattern. 1 Introduction & Related Work The term transport is used to describe a variety of phenomena from different con- texts, and occurring across varying spatial scales. In the field of transportation, transport primarily refers to the movement of people or goods within an infras- tructure such as a road or airline network. In physical and chemical systems, transport refers to the movement of molecules or other physical matter, energy, or momentum, from one system to another. In computer networking, packets of digital information are transported. We also talk about humans transporting thoughts, ideas, and opinions through communication. In all these cases, there is movement of some entity via a transport mechanism from one place to another. The Semantic Transport Ontology Design Pattern (ODP) is designed as a basic, extensible foundation for modelling transport concepts and relations in an ontology [4, 7]. In the geosciences, transport is a prevalent concept. Within the popular Semantic Web for Earth and Environmental Terminology (SWEET) ontology7 [8], for example, there are at least 22 different concepts that reference “transport” in some form. By using it as a template for describing the relations between the mechanisms and entities involved in transport events, the Semantic Transport ODP can enrich descriptions of scientific data sets to aid in their interoperability, and re-use; as well as for data mining. 7 http://sweet.jpl.nasa.gov/ontology/ The Semantic Transport ODP is designed to be compatible with other ODPs generated during the GeoVoCamp8 series of workshops. Previous workshops fo- cused on a range of geo-spatial topics from cartographic map scaling [2] to se- mantic trajectories [5]. Of particular note, the Semantic Transport ODP can operate in tandem with the the Semantic Trajectory pattern [5] as the former describes the entity and energy of transport, and the latter describes the path along which the transport occurred. The Semantic Transport pattern is also conceptually related to the proposed Move ontology design pattern9 that is derived from the CIDOC model [3]. The Semantic Transport pattern, however, decouples the source energy from the en- tity being displaced while capturing their interdependence. 2 Transport Pattern In this section we present the core elements of the Semantic Transport pattern, and describe how it can be extended to cover two different types of transport: active and passive. We focus on applying the Semantic Transport ODP in the context of physical systems, though it may also be useful for other domains, e.g. describing cultural transmission. 2.1 Core elements The Semantic Transport pattern consists of three core concepts: Event, Entity, and Mechanism (see Manchester OWL syntax following this paragraph). The TransportEvent acts as the top level concept for the pattern. A TransportEvent describes a specific transport phenomenon, as movement of some mass or en- ergy (measurable entity) from one location to another, based on a common and persistent frame of reference. Induction of the mass or energy movement can arise from the transported entity itself, or from external sources. The Trans- portEvent thus has two main parts, TransportEntity and TransportMechanism. The TransportEntity concept represents the identity of the circumscribed por- tion of energy or mass that is moved. The TransportMechanism concept captures the nature of the source that acts upon the TransportEntity, and thus induces a TransportEvent. Class: TransportEvent TransportEvent SubClassOf owl:Thing TransportEntity SubClassOf partOf some TransportEvent TransportMechanim SubClassOf partOf some TransportEvent Class: TransportMechanim TransportMechanim SubClassOf owl:Thing TransportMechanim SubClassOf partOf some TransportEvent 8 http://vocamp.org/wiki/GeoVoCampSB2013 9 http://ontologydesignpatterns.org/wiki/Submissions:Move Class: TransportEntity TransportEntity SubClassOf owl:Thing TransportEntity SubClassOf partOf some TransportEvent The TransportEvent has one top level property, the referenceFrame (see Manchester OWL syntax following this paragraph). The referenceFrame pro- vides context to the pattern by specifying spatial and temporal qualities of any associated observations via the specification of time and location information associated with the TransportEvent. As time and location can be fixed or rela- tive, abstracting the property types serves to facilitate semantic interoperability between disparate data entities. ObjectProperty: referenceFrame referenceFrame Domain TransportEvent referenceFrame Range TransportEvent It is worth noting that even if the transported entities were to return somehow to their exact place of departure, they still participated in a TransportEvent, even if their initial and final locations result in no net change in location. 2.2 Extending the Transport Pattern Figure 1 illustrates the Semantic Transport pattern along with a few logical ex- tensions to illustrate how the pattern might be used. The pattern constructs are depicted in figure 1 using black text in translucent shapes. Further, all classes in the figure are represented using an oval, and each square box delineates a property. In most cases there will be interest in a more specific description of the event. These additional aspects are accommodated by extending the pattern through subclassing of the existing TransportMechanism and TransportEntity concepts, while adding properties to the referenceFrame (Figure 1). These ad- ditional pragmatic components are included in figure 1. Class symbols with red text are examples of the nomenclature that may be used as a TransportMecha- nism. Sub-properties with blue text illustrate examples of modules that may be described as part of the constructs comprising the referenceFrame. The TransportMechanism can be usefully subclassed into specialised topics, such as in the case of involving disjoint classes such as PassiveTransport and ActiveTransport. Classic examples of PassiveTransport include, e.g. diffusion or osmosis, while ActiveTransport would include, e.g. ATP pumps or air travel [1]. The referenceFrame is a non-trivial property. Geoscience phenomena often exhibit unique statistical signatures as mechanical, chemical, and biological pro- cesses work in tandem, or asynchronously, through time as they tend toward equilibrium [6]. This property is significant in that it serves to preserve the spa- tiotemporal information necessary to maintain a consistent granularity of context throughout the pattern. Fig. 1. Transport Pattern extended. Each class in the figure is represented by an oval, and each square box delineates a property. The Semantic Transport pattern constructs are depicted using black text in translucent shapes. Classes using red text are logi- cal subclasses, while sub-properties are styled using blue text. Three equivalent class relations are represented on the right of the figure, each one illustrating how the refer- enceFrame can connect to other established schemas. Further, an observation (as it relates to the referenceFrame in figure 1) is considered semantically equivalent to other well established geoscience explica- tions, namely the observation entities associated with the Sensor Observation Service10 (SOS) and the Geography Markup Language11 (GML), as well as the concept of a “fix” in the Semantic Trajectory pattern [5]. Of course, further subclassing will likely be necessary for this pattern to connect, and be useful, to much of the disparate data available to geoscientists. The current framework is complete and extendable. By creating logical semantic equivalences to constructs already used throughout the domain, the Semantic Transport pattern can be a powerful module when mining and filtering large data stores. 3 Summary and Future Work In this paper we presented an ontology design pattern to describe transport phe- nomena. The core Semantic Transport ODP is deliberately simplified to essential elements, to be applicable to a wide range of use cases in the physical sciences. We described how the Transport pattern can be extended for Active and Passive transport and illustrate briefly how it might be used to interoperate over large disparate geoscience data. 10 http://www.opengeospatial.org/standards/sos 11 http://www.opengeospatial.org/standards/gml Next steps will involve how the ODP can be filled out to describe data for a variety of use cases, as well as application and usability testing. An important future extension to the pattern for application to the physical sciences will in- clude explicating the relationship between concepts such as system and energy input (in terms of entropy of the system). References 1. Berg, H.C.: Random walks in biology. Princeton University Press (1993) 2. Carral, D., Scheider, S., Janowicz, K., Vardeman, C., Krisnadhi, A., Hitzler, P.: An ontology design pattern for cartographic map scaling. In: Cimiano, P., Corcho, O., Presutti, V., Hollink, L., Rudolph, S. (eds.) The Semantic Web: Semantics and Big Data, Lecture Notes in Computer Science, vol. 7882, pp. 76–93. Springer Berlin Heidelberg (Jan 2013) 3. Doerr, M.: The cidoc conceptual reference module: an ontological approach to se- mantic interoperability of metadata. AI magazine 24(3), 75 (2003) 4. Gangemi, A.: Ontology design patterns for semantic web content. In: The Semantic Web - ISWC 2005. Lecture Notes in Computer Science, vol. 3729. Springer (2005) 5. Hu, Y., Janowicz, K., Carral, D., Scheider, S., Kuhn, W., Berg-Cross, G., Hitzler, P., Dean, M.: A geo-ontology design pattern for semantic trajectories. In: COSIT 2013. Springer (2013) 6. Kastens, K., Manduca, C., Cervato, C., Frodeman, R., Goodwin, C., Liben, L., Mogk, D., Spangler, T., Stillings, N., Titus, S.: How geoscientists think and learn. Eos. Trans. AGU 90(31), 265–266 (2009) 7. Presutti, V., Gangemi, A.: Content ontology design patterns as practical building blocks for web ontologies. In: Li, Q., Spaccapietra, S., Yu, E.S.K., Olivé, A. (eds.) ER. Lecture Notes in Computer Science, vol. 5231, pp. 128–141. Springer (2008) 8. Raskin, R.G., Pan, M.J.: Knowledge representation in the semantic web for earth and environmental terminology (SWEET). Computers & Geosciences 31(9), 1119– 1125 (2005)