GeoData (CTU + LRI)

Application domain KDD in domain of geographic databases
Source data extracted from a map manually
Dataset size 105 facts
Data format plain LP
Systems used WiM
References [1]

The dataset contains a simplified description of rivers, roads, rail-ways as well as woods/forests and buildings (one fact per object) corresponding to real-world topographic data. Each object is described by its geometry using a 2-dimensional projection, some additional characteristics are introduced when neccessary, e.g., for the roads and railways. The applied modification maintains the essential geographical features of the original data with clear geometrical interpretation (e.g. line intersections, the size of the considered area etc.).

The background knowledge utilizes those basic elementary concepts, which are frequently used to describe contents of a map, e.g. road and its type, water stream. The experiments have been conducted with intention to obtain an explicite characterization of some more complex object implicitely identified through examples. One of successful attempts induced e.g. a simple set of rules distinguishing a forest from a wood.


  1. L.Popelynsky, 'Knowledge Discovery in Spatial Data by Means of ILP' in Principles of Data Mining and Knowledge Discovery, Proc. of 2nd European Symposium PKDD'98 (Zytkow J.M., Quafafou M., eds.), Nantes France 1998. LNCS 1510, Springer-Verlag 1998.

back to index