Co-location Pattern Mining Under the Spatial Structure Constraint.

Most methods to find spatial co-location patterns (subsets of object features that are geographically close to one another) employ standard proximity measures (e.g. Euclidean distance). But for some applications, these measures do not work well since the spatial structure is not considered. This article proposes CSS-Miner, a co-location pattern mining approach under the spatial structure constraint. In this case, the street network of a city is used as a constraint. CSS-Miner has been applied to two real datasets with different points of interest.