Antony Galton

The purpose of this tutorial is to provide an overview of current and recent thinking on the description, classification and representation of dynamic phenomena in the geographical domain. The tutorial should provide sufficient background for newcomers to the field to be able to appreciate research contributions in this area, as represented in conferences such as COSIT, and journals such as IJGIS. The tutorial is designed to meet the needs of PhD students and post-doctoral researchers in GI Science, spatio-temporal reasoning, and related fields. Some familiarity with formal notations such as logic and set theory would be useful (particularly for Session 3), as well as with methods of representing static information in GIS.

We will look at the variety of processes and events that must be considered in describing dynamic geographical phenomena, different ways of classifying them, how they fit into high-level representation schemes for geographical information, and a variety of formal approaches to the representation and modelling of such phenomena.

The tutorial will comprise four 45-minute sessions with a half-hour break between the second and third (total duration 3 hrs 30 minutes). The programme will be as follows:

• Session 1: Scope [pdf slides]
  Survey of kinds of processes and events of interest to GI Science.
  Terminological matters: the relationship between 'process' and 'event'; classic taxonomies (e.g., Vendler, Mourelatos).
  Varieties of process - steady-state, cyclic, cumulative, irregular, etc. Importance of perspective and granularity.
• Session 2: Handling processes and events in GIScience [pdf slides]
  Degrees of temporal involvement in GIScience: snapshots, object lifelines, explicit representation of states, events, and processes (following Worboys and Duckham 2004).
  Three-dimensionalist vs four-dimensionalist approaches, SNAP/SPAN, other ontological considerations.
• Session 3: Formal Treatments [pdf slides]
  Different approaches to the logic of time: modal vs first-order approaches, and within the latter, various degrees of reification. Instants vs intervals.
  Other approaches to temporal representation, e.g., interval calculus, event calculus.
• Session 4: Representation and modelling [pdf slides]
  Information Systems vs Modelling Systems ("factual" vs "fictional" time).
  The importance of causation and kindred notions such as prevention, enablement, perpetuation.
  Eulerian vs Lagrangian models; map algebra (applied to snapshots rather than layers), cellular automata, agent-based models, process calculi, etc.
  

In addition to the session slides, which can be downloaded from the links above, you can also download the tutorial notes from this page.