Discovering Geo-referenced Frequent Patterns in Uncertain Geo-referenced Transactional Databases

An uncertain geo-referenced transactional database represents the probabilistic data produced by stationary spatial objects observing a particular phenomenon over time. Useful patterns that can empower the users to achieve socio-economic development lie hidden in this database. Finding these patterns is challenging as the existing frequent pattern mining studies ignore the spatial information of the items in a database. This paper proposes a generic model of Geo-referenced Frequent Patterns (GFPs) that may exist in an uncertain geo-referenced transactional database. This paper also introduces two new upper-bound constraints, namely "neighborhood-aware prefix item camp" and "neighborhood-aware expected support", to effectively reduce the search space and the computational cost of finding the desired patterns. An efficient neighborhood-aware pattern-growth algorithm has also been presented in this paper to find all GFPs in a database. Experimental results demonstrate that our algorithm is efficient.