Event-based modeling for geographic domains
Investigator: Kathleen Stewart Hornsby
Modeling time-varying phenomena is a key topic for many researchers in geographic information science and related fields. Most of this research, however, has focused on modeling entities as objects. Modeling the dynamic aspects of a domain as events provides semantics that are missing from approaches based on a solely object-centered approach.
The creation of more realistic spatio-temporal data models relies on an understanding of how entities and events are related. For example, entities often experience more than one event, such as when a vehicle encounters a traffic jam, turns onto a side street to avoid the jam, and then rejoins traffic at a point further along the roadway. Describing such a set of occurrents forms an ordering that can be represented as a sequence of events. Just as humans perceive the passage of time as a series of succeeding occurrents, the movement of an object may be described by a sequence of events. As part of this research, we model the events experienced by entities in a geospatial domain (e.g., moving vehicles or ships), and from an analysis of these events, we abstract meaningful information about, for example, the movement of entities. The inclusion of events in geographic data models provides a foundation for distinguishing a wide range of semantics of movement based on patterns of events.
In a geographic domain, an object’s movement through the domain can be represented as an event sequence referred to as a transit T, 
, where ID is the identity of the moving object. A transit models all the movement-related events experienced by a particular object during a traversal of the domain. Formally, a transit is defined as: . As part of this research, we examine the semantics of transits, highlighting the different combinations of events that
can lead to different types of movements.
In additional research, we examine the changes that result from events, i.e., impacts of events. A fire event in a nature reserve, for example, burns trees and bushes resulting in an abrupt decrease in vegetation height in the reserve. A traffic accident on an interstate highway can result in an impact, such as a traffic jam, or lane closures, or both. As part of this research we develop a qualitative classification of impacts in order to reason about events and their impacts. The underlying conceptual model distinguishes between immediate and delayed impacts. Based on this distinction, a set of basic types of impacts are differentiated, in particular abrupt and gradual impacts. We analyze how the temporal relation between an event and an impact can be used to capture combinations of impact types, called evolving impacts. To link event-impact relations spatially, this work introduces the concept of qualitative impact maps that represent the extent of an impact type. The combination of qualitative impact maps with event-impact inference rules enables the identification of events that are likely to underlie these impacts. The application potential of this approach is demonstrated via a case study based on vegetation change and event data for a nature reserve near Melbourne, Australia.
This research is supported in part by the National Geospatial-Intelligence Agency under grants NMA201-00-1-2009 and HM1582-05-1-2039.
Related publications
K. Stewart Hornsby and M. Yuan (Eds.)(2008) Understanding dynamics of geographic domains, CRC Press, New York, NY, in preparation.
M. Yuan and K. Stewart Hornsby(2007) A research agenda for the computation and visualization for the understanding dynamics of geographic domains, CRC Press, New York, NY, in preparation.
K. Hornsby and S. Cole (2007) Modeling moving geospatial objects from an event-based perspective,Transactions in GIS, 11(4): 225-243.
I. Mau, K. Hornsby and I. Bishop (2007) Modeling geospatial events and impacts through qualitative change, Proceedings of the International Conference on Spatial Cognition, Spatial Cognition 2006, Lecture Notes in Artifical Intelligence, 4387, Springer, Berlin, pp. 156-174.
S. Cole and K. Hornsby (2005) Modeling Noteworthy Events in a Geospatial Domain, Proceedings of the First International Conference on Geospatial Semantics, GeoS 2005, Lecture Notes in Computer Science, 3799, Springer, Berlin, pp. 78-92.
S. Hall and K. Hornsby (2005) Ordering events for dynamic geospatial domains, in A. Cohn and D. Mark (eds.) Proceedings of the Seventh International Conference on Spatial Information Theory, COSIT 2005, Lecture Notes in Computer Science 3693, Springer, Berlin, pp. 330-346.
K. Hornsby (2004) Retrieving event-based semantics from images, Proceedings of the IEEE Sixth International Symposium on Multimedia Software Engineering, 13-15 December 2004, Miami, FL, IEEE Computer Society, pp. 529-536.
K. Hornsby and M. Worboys (Eds.) (2004) Special issue on event-oriented approaches for geographic information science, Spatial Cognition and Computation, 4(1).
M. Worboys and K. Hornsby (2004) From objects to events: GEM, the geospatial event model, In M. Egenhofer, C. Freksa, H. Miller (Eds.) Proceeding of GIScience 2004, Lecture Notes in Computer Science, 3234, Springer, Berlin, pp. 327-343.
K. Hornsby and S. Hall (2004) Generating linear orders of text-based events, Proceedings of the Workshop on Computational Lexical Semantics, Human Language Technology Conference 2004 (HLT/NAACL’04), Boston, MA. pp 92-99.
