Mining local association patterns from spatial dataset
Zongyao Sha, Xiaolei Li · 2010 Seventh International Conference on Fuzzy Systems and Knowledge Discovery · 2010
This paper proposed a model and algorithm to mine local association rules from existing spatial dataset while fully taking the fact that spatial heterogeneity may widely exist in reality. The essential part of the model is the calculation localized measure of association strength (LMAS) which is used to quantify local association patterns. Spatial association relations are specifically defined as spatial relations which are modeled by DE-9IM model. We proposed mining algorithm for discovering local association patterns from spatial dataset. The proposed algorithm extracts reference and target objects that have potential association patterns and processes LMAS for each object in the reference objects for any interested spatial relation. Therefore, the output of the algorithm is a LMAS distribution map that reflects association strength variations over the study region. Spatial interpolation for LMAS is suggested to create a continuous LMAS distribution which can be used to explore “hot” spots that demonstrate strong association patterns. This proposed model and algorithm was applied in a ecological system research.