Efficient Discovery of Top-K Sequential Patterns in Event-Based Spatia-Temporal Data

We consider the problem of discovering sequential patterns from event-based spatio-temporal data. The dataset is described by a set of event types and their instances. Based on the given dataset, the task is to discover all significant sequential patterns denoting the attraction relation between event types occurring in a pattern. Already proposed algorithms discover all significant sequential patterns based on the significance threshold, which minimal value is given by an expert. Due to the nature of described data and complexity of discovered patterns, it may be very difficult to provide reasonable value of significance threshold. We consider the problem of effective discovering K most important patterns in a given dataset (that is, discovering top-K patterns). We propose algorithms for unlimited memory environments. Developed algorithms have been verified using synthetic and real datasets.