Discovering Sequential Patterns in Event-Based Spatio-Temporal Data by Means of Microclustering - Extended Report

TL;DR

This paper introduces a method for efficiently discovering sequential patterns in event-based spatio-temporal data by reformulating existing approaches and leveraging microclustering techniques.

Contribution

It presents a novel reformulation of sequential pattern discovery in spatio-temporal data using microclustering to improve efficiency and effectiveness.

Findings

Effective discovery of sequential patterns in spatio-temporal data

Reformulated approach enhances computational efficiency

Patterns reveal attraction relations between event types

Abstract

In the paper, we consider the problem of discovering sequential patterns from event-based spatio-temporal data. The problem is defined as follows: for a set of event types $F$ and for a dataset of events instances $D$ (where each instance in $D$ denotes an occurrence of a particular event type in considered spatio-temporal space), discover all sequential patterns defining the following relation between any event types participating in a particular pattern. The following relation $\rightarrow$ between any two event types, denotes the fact that instances of the first event type attract in their spatial proximity and in considered temporal window afterwards occurrences of instances of the second event type. In the article, the notion of sequential pattern in event-based spatio-temporal data has been defined and the already proposed approach to discovering sequential pattern has been reformulated. We show, it is possible to efficiently and effectively discover sequential patterns in event-based spatio-temporal data.