Ian Anderson
Research Interests: Summary
Models of space can typically be classified as either topological (qualitative) or, more commonly, as coordinate based (quantitative). Quantitative models generally take a geometric view of space with positional information supplied by location services using Euclidean or spherical coordinate systems. Coordinate tuples are processed by the application and behaviour is updated to reflect the new location information. In contrast, topological or symbolic models manage space in a qualitative manner with positional information mapped to human abstractions of physical places, usually in the form of spatial zones. The relationships between zones form a topology often expressed as a graph. Application behaviour varies depending upon the symbolic representation of space (zone) that the user is currently located in. When constructing a symbolic model of the spatial environment developers must define spatial zones within the constraints of the underlying sources of positional information. For example, it is not possible to create zones with a physical coverage area that is finer than the granularity of the data produced by the positioning services. I am currently investigating methods of automating the generation of optimal qualitative models of the application environment in an unsupervised manner given the positional measurement and environmental constraints. |