Temporal Dynamics and Success Patterns of Online Petitions: A Time-Series Clustering Approach
Mate Kovacs, Daniil Buryakov, Didier Gohourou, Uwe Serdült · 2024
Online petition systems worldwide enable individuals to express about public life issues. Usually, these systems require a certain threshold of digital endorsements to be met within a given timeline before a petition can reach the authorities. In the Taiwan’s JOIN platform the requirement is 5000 signatures within 60 days. However, the dynamics of endorsements for online petitions can vary greatly over time. An analysis of all admitted online petitions since 2015, applying Dynamic Time Warping and k-means clustering, produced three clearly distinct clusters. These clusters correlate with the varying success rate of online petitions as well as time required to achieve the signature threshold. Out of three clusters, two accounted for 85% of successful petitions. The study aims to identify factors that explain the observed patterns in successful petitions, specifically focusing on the topics of the petitions and the days they were initiated and completed. The implemented approach not only improve the operational understanding of petition platforms but also offers tools to potentially detect anomalies that may suggest manipulative activities. Additionally, the methods developed could facilitate comparisons among different platforms.