Recurrence dynamics of Internet access user activity

Abstract

Internet has been an important tool in connecting, interacting, and garnering information in this digital age. This study is on the behavior of Internet access requests through the Diliman Network (DilNet) of the University of the Philippines Diliman (UPD) analyzed using recurrence plots (RP) and recurrence quantification analysis (RQA) measure: recurrence rate r. A six-day data set was investigated using RP with parameters: time delay τ, epsilon ε, and embedding dimension M chosen appropriately. The resulting recurrence plots show disrupted typology with clusters of diagonal and vertical lines indicating activity aggregation. The recurrence rate is shown to decrease according to power law r ~ ε^α with α that is sensitive to M. The deviation of obtained α values is discussed in relation to the activity being constant or intermittent. The computed values of r for the same M at different days also reflect high activities, but vary in interpretation depending on the corresponding α-values. The results were related to the expected regular activities of DilNet users.