130 likes | 141 Views
Geospatial Ontologies - A View from NGA. Dan Adams Senior Research Scientist NGA Office of Basic and Applied Research Daniel.R.Adams@nga.mil. Standards, Canon, Opinions, and Disclaimers. NGA presently has no standard geospatial ontology/ontologies
E N D
Geospatial Ontologies -A View from NGA Dan Adams Senior Research Scientist NGA Office of Basic and Applied Research Daniel.R.Adams@nga.mil
Standards, Canon, Opinions, and Disclaimers • NGA presently has no standard geospatial ontology/ontologies • NGA presently has no policies for the development and deployment of ontologies • But interest is growing… • What follows is the view from an Applied Research perspective
Ontologies: Of What … and Why • Domain to Data • Discovery • Retrieval • Integration • Data to Data • Integration • Knowledge↔Data • Features • Names • Topography • Geography (physical, political, …) • Topology • Mereology • Provenance • Domains
Top Down, Bottom Up or Middle Out? • No apparent interest in, or traction for, use of upper ontologies • Efforts thus far have been Bottom Up… • University research in codification of things observable on the ground • Promotion of Feature Catalog to an ontology • Ontology layers over data source such as airfields (DAFIF™), vertical obstructions (DVOD), GeoNames, etc.
Top Down, Bottom Up or Middle Out?(cont) • And Middle Out • Narrow domains with hand-crafted ontologies bound to lower ontologies • Data ontologies bound upward (to Cyc) • Geospatial Semantic Web research program (started late in 2004) has been primarily working with lower and middle ontologies
Why is NGA Interested in Ontologies? • Organization of Information • “Smart” Taxonomies • Information Discovery • Integration of Data Sources • Many databases • Overlapping, sometimes inconsistent, attributes • Knowledge Capture and Exploitation • More than just ontologies Greatest Hype Risk
Ontology Development Activities at NGA • NURI grant for landscape-based feature classification taxonomy - 2003 • Extracting ontologies in OWL format from Cyc (after enriching the geospatial knowledge in Cyc) - 2005 • Concept Maps vice Ontologies for capturing domain knowledge - 2006 • And possibly binding concept maps to ontologies • “Mini-” ontologies for research prototypes
Where are Ontologies Needed Now?(Feature level) • (Semantic) integration of databases • GGMA is the most visible project (but NGA has a lot of databases…) • Which ontologies will facilitate this? • Discovery and Sharing of relevant data • Within NGA • Between agencies (e.g., DIA, DHS, USGS) • Augment – not replace – metadata
Where will Ontologies be Needed Next?(Add concept and process level) • Enriched analytical products • Move beyond words and embedded pictures • Smarter analytical environments • Assist with process of analysis
Where the Heavy Lifting is Needed • Build a good ontology of geospatial features to support data integration right now • Try to use what’s already been done • Leave room to grow and adapt • Anticipate and prepare for overlapping ontologies • Keep the metadata distinct from the ontology • Check GIS and RDBMS biases at the door
Phrases You Never Expected to Hear… • “There simply aren’t enough good philosophers to go around.” -- Mark Musen (attributed to Barry Smith?) • “We’re going to have to start taking this semantic web stuff seriously.” – Michael Witbrock
Know the Earth…Show the Way UNCLASSIFIED NATIONAL GEOSPATIAL-INTELLIGENCE AGENCY UNCLASSIFIED