Two Stage Knowledge Discovery for Spatio-temporal Radio-emission Data

In this paper, we introduce a method to examine and interpret spatio-temporal radio emission datasets. The goal is to find communication patterns in the data in respect to spatial, temporal, and frequency based attributes. The chosen approach is a combination of two different AI-methods. First a clustering algorithm groups spatially close data points to potential emitters. In a second step a model-based constraint solving technique is applied to find relationships between the identified emitters. The used models describe rules of the communications that are to be found. This guarantees a flexible search for different kinds of communication.