Topological Reasoner

an interactive demonstion for a simple geographic example

by Brandon Bennett

The map image oposite was originally represented in a simple vector format, in terms of triangulations of a set of land types. RCC topological relations between the land types have been precomputed from this data. This gives the following qualitative representation of the island in terms of RCC relations:

   ec(  sea,     town   )
   ec(  sea,     plain  )
   dc(  sea,     forest )
   ec(  sea,     desert )
  tpp(  town,    plain  )
   dc(  town,    forest )
   dc(  town,    desert )
ntpp1(  plain,   forest )
   ec(  plain,   desert )
   dc(  forest,  desert )
The following qualitative facts which partially constrain the locations of the habitats of various animals are also known:
   pp(  buffalo,  plain  )
   dc(  buffalo,  town   )
  tpp(  monkey,   forest )
    o(  lizard,   desert )
   p1(  frog,     forest )
   po(  frog,     plain  )
equal(  whale,    sea    )
Using an encoding into intuitionistic logic we have implemented a reasoning algorithm which can test consequences of the combined constraints on the terrain types and the animal habitats.