1. Semantic Evolution of Geospatial Web Services: Use Cases and Experiments in the Geospatial Semantic Web Joshua Lieberman, Todd Pehle, Mike Dean Traverse Technologies, Inc., Northrop Grumman Information Technology / TASC, BBN Technologies

2. © 2004 Traverse Technologies Inc. Overview  Geospatial information and Opengeospatial Web Services  Geospatial Semantic Web, an interoperability experiment  Semantic challenges on the Spatial Web

3. © 2004 Traverse Technologies Inc. What are OGC and OWS ?  “ The Open Geospatial Consortium, Inc. (OGC) is a non-profit, international, voluntary consensus standards organization that is leading the development of standards for geospatial and location based services”  “Opengeospatial Web Services” (OWS) - OGC has been developing for some time specifications for a suite of Web services (sensu lato) and associated encodings to expose geospatial content and operations from distributed content repositories to remote clients across diverse platforms:  GML - geographic markup language (an information model and XML schema) for encoding features (geometric representations of geography).  Web Feature Service - service providing access to collections of features  Web Map Service - service providing access to map layers (cartographically rendered features and images)  Catalog Service / Web - service supporting (spatial) discovery of geospatial datasets and services  Several other associated specifications

4. © 2004 Traverse Technologies Inc. 1) GSW Background  Geospatial Semantic Web : Use of Semantic Web technologies to discover and reason on geospatial information (UCGIS, Egenhofer, Sheth, etc.)  GSW broad research activity sponsored by National Geospatial Intelligence Agency (NGA), undertaken by a number of investigators  Interoperability experiment : an Open Geospatial Consortium(OGC) - sanctioned member collaboration to test or refine OWS specifications  This “ GSW IE ”: activity proposed by NGA, NGIT/TASC, and BBN to test and refine OGC(+) specifications within a scenario for geospatial query with formal semantics:  Web Feature Service (WFS) and Filter Encoding (FE)  Geography Markup Language (GML)  ISO 19115 / 19119 / 1910n / FGDC feature metadata (ISO)  Other initial participants: SCO, Jaume I, Muenster, Galdos, GMU, …

5. © 2004 Traverse Technologies Inc. Drilling Down:  The Geospatial part  Maps and map visualization  Features and feature geometries  Geographic and other relationships  The Web part  Distributed data - “maintain locally / access globally”  Shared services, loosely or tightly coupled to geodata  Interoperability between technologies, vendors, architectures  The Semantic part  Interoperability between communities and domains  Softer software  Automated reasoning and inference

6. © 2004 Traverse Technologies Inc. Geospatial Reasoning: 2-D and Beyond  Coordinate relationships  Scale significance  Coordinate reference systems  Topological relationships  Network  Overlay  Spatial inference  Proximity  Continuity  Representation  Dimensionality  Temporality

7. © 2004 Traverse Technologies Inc. The Web Changes Everything (Geospatial)  Global communities for local geography  Distributed information networks  Premium on interoperability  The GIS dialtone  Maintain locally, access globally  Currency is the currency  (non-GIS) barbarians are at the (GIS) gate

8. © 2004 Traverse Technologies Inc. Role of interoperability / opening the temple  Focuses on sustained operability - today and the next day  Permits separation of concerns  Supports information portability  Allows component interchangeability  Contributes to transparency, testability, and trust  Layers of interoperability build on one another  Stable syntax promotes shared semantics / understanding  Standards are necessary but not sufficient for interoperability

9. © 2004 Traverse Technologies Inc. 2) Interoperability experiment: goals  Exercise current semantic technology in a geospatial realm  Demonstrate an end-to-end geospatial semantic query  Utilize multiple ontologies for Geointel operations  Develop OGC service descriptions with formal semantics (e.g. OWL-S description for Web Feature Service)  Develop and test Semantic Web Services interface / role for OGC services  Enhance interoperability in a distributed, heterogeneous world, or at least identify the problems

10. © 2004 Traverse Technologies Inc. “Typical” Geospatial Query (Intelligence / Logistics Domain) “Which airfields within 500 miles of Kandahar support C5A aircraft?” Aero Feature or Geo Feature? Buffer or proximity? Statutory or Nautical? Straight-line or driving? Coordinate system? Afghanistan? Centroid or outline? What does this mean? Feature property or non-spatial information?

11. © 2004 Traverse Technologies Inc. Sequence of Experimental Tasks Link Ontologies into Knowledgebase Generate and Visualize OWS (WFS) Queries Request Remote Service Descriptions Process Queries Through Knowledgebase Compose Queries and Query Templates Generate and Distribute Sub-queries Identify and Build Ontologies for Geospatial / GeoIntel Domains

12. © 2004 Traverse Technologies Inc. Multiple GSW Ontology Components GeoIntel Problem Domain Ontology Base Geospatial Ontology NGA Feature Ontology OGC Services Ontology Other Base Ontologies

13. © 2004 Traverse Technologies Inc. Initial ECDM Selections for Experiment Airport Runway Taxiway Apron ObstructionThreshold Route Plane Itinerary Repair Weather Nav Aids Service Fuel Lighting VOR NDB ILS MLS TACAN

14. © 2004 Traverse Technologies Inc. “Typical” GSW Query Stack  Query / Visualization Client  Query Translator  Query Processor  Graph Store / Inferencer  Geospatial Inferencer  Remote WFS harvester  Remote WFS translator  Remote WFS  Do CSW query, then WFS query  Translate GetRecord to semantic query  Process semantic query  Query knowledgebase  Resolve geospatial relations  Add WFS descriptions to knowledgebase  Translate between GML / XML & OWL  Provide GML features through WFS interface Standardized Web Services interfaces can be (or have been) defined between any two of these stack elements Element Function

15. © 2004 Traverse Technologies Inc. Model Query “Plan” Question Template Query Rules & Artifacts Knowledge Base Reasoning & Inference Domain Ontology Ontologies Remote WFS Get Feature Local Ontologies Visualizer Map Knowledge Server Knowledge Server Sub-query Service Response Query Client Visualization Client

16. © 2004 Traverse Technologies Inc. GEOINT Query Plan Has…  Concepts/Relationships, e.g. OWL ontology elements  Rules, e.g. RuleML (SWRL)  Completion Criteria, e.g. SeRQL query elements  (precondition) Inference-based knowledge refinement  (precondition) Traversal of geospatial relationships  (precondition) Access to remote services through semantic service descriptions Question Template Query Rules Domain Ontology Query Client

17. © 2004 Traverse Technologies Inc. (Some) Technology Options ArcGIS / Gaia Semantic Feature Visualizer Plugin Semantic Query Plugin DbSAILWebSAIL GeoSAIL Query Layer (SeRQL) HTTP API Sesame Processing Framework DamlDB WFS Java Topology Services Templates REP APIGRAPH API Oracle 10g

18. © 2004 Traverse Technologies Inc. OWL-S Service Description Components and Questions Type of Service Themes of Content Provider / business terms Content Description Service Bindings / Messages Bound Parameters Service Quality Smart Service Consumption Service Composition Service Profile Service Grounding Service Model Feature Schema Content Domain Feature Individuals ?

19. © 2004 Traverse Technologies Inc. 3) Semantic service description: the missing bits  > description  Profile (e.g. ISO 19115)  Grounding (bound parameters)  Model (e.g. GML, ISO 19110)  Interface for query of service description  Progressive generalization  Discovery  Binding  Domain  Dictionary  Individual / Instance  Description of service self- description

20. © 2004 Traverse Technologies Inc. Observations  The geospatial realm has well-developed information and service models, but mainly implicit semantics  Within the geospatial realm are many communities with only partially shared vocabularies.  Gradual adaptation of the existing data and infrastructure is essential  Geospatial content is (necessarily) scale-dependent, distributed, heterogeneous, and dynamic - a challenge for description / generalization  Geospatial services are typically tightly coupled to content  Resolution / traversal of geospatial relationships is a computational challenge  Semantic Web Services are essential for opening up the “geospatial temple cult” but must avoid exchange of one cult for another