Mining Co-locations under Uncertainty

Zhi Liu, Yan Huang

Abstract: A co-location pattern is a subset of spatial features whose events tend to occur together in proximity. Under uncertainty (imprecise or noisy locations), over-counting across possible worlds is a challenge. We define a probabilistic participation index based on the possible-worlds model and prove a lemma enabling instance-centric counting that avoids over-counting while matching possible-worlds results. Leveraging this, we develop efficient mining algorithms for uncertain spatial data.