Dangoron: Network Construction on Large-scale Time Series Data across Sliding Windows

Abstract

In complex networks, the dynamics of systems are represented through the interactions of a set of anomalous time series. A crucial problem to consider is the computation of correlations between highly correlated pairs of time series across sliding windows. The efficient calculation and updating of the correlation matrix, considering user-defined sliding periods and thresholds, are vital for facilitating large-scale time series network dynamics analysis. We present Dangoron, a framework meticulously designed for the efficient identification of highly correlated pairs of time series over sliding windows and the precise computation of their respective correlations. Dangoron predicts dynamic correlations across sliding windows and prunes unrelated time series, thereby yielding a performance at least an order of magnitude faster than a baseline approach. Additionally, we introduce Tomborg, the first benchmark specifically developed to address the challenge of correlation matrix computation in the context of time series analysis. This benchmark serves as a robust foundation for future research in this domain.