Lunch talk on Mar. 20th, 2019
High Quality Co-location Pattern Mining
Speaker: Yuan Fang (Yunnan University)
Venue: Gewu 4410
Time: 12:30 PM, Wednesday, 20th March, 2019
Abstract: Spatial co-location pattern mining technique aims to discover the subsets of spatial features whose instances are frequently located together in geographic neighborhoods. The traditional co-location mining framework takes the prevalence as sole interestingness measure. Such framework produces numerous co-location patterns and only uses the co-existence intensity of features as pattern information, which makes the mining results hard for users to understand and apply in specific applications. In order to acquire a kind of concise, targeted and interpretable co-location patterns, we consider the interaction of instances, features and co-location patterns in co-location pattern mining process simultaneously, and define the high-quality co-location patterns (HQCPs). Moreover, we develop efficient algorithms and a series optimized strategies to discover three kinds of HQCPs: the co-location patterns with dominant features, dominant co-location patterns and combined co-location patterns. The experimental results on vegetation data sets show that our methods which evaluate and analyze the co-location patterns on multiple perspectives can find a collection of co-location patterns with high quality to guide the users in practice.