Dealing effectively with overwhelming masses of spatial data in the twenty-first century will require more than improved technologies for faster computers or parallel algorithms. Effectiveness incorporates but should not be confused with efficiency; the more fundamental challenge remains ultimately the selection of relevant data for further attention or processing by humans or computers. This paper defines a relevance filter as a mechanism interposed between an extensive source of input data and a human or computer agent that seeks to make use of some subset of the data for a particular analytical purpose. Consideration of the central principles for the design of such relevance filters raises a number of theoretical and practical research questions related to spatial information, and unifies several surprisingly diverse lines of research.

Catherine Dibble 27 June 2003