Modeling Collective Behavior of Posting Microblog by Stochastic Differential Equation with Jump

TL;DR

This paper introduces a stochastic differential equation model with time-dependent parameters and jump processes to better understand and predict collective microblogging behavior, including periodic patterns and sudden spikes, on Sina Weibo.

Contribution

It presents a novel SDE model incorporating time-varying parameters and compound Poisson jumps, improving prediction accuracy of microblogging activity over existing models.

Findings

Model captures periodic and jump behaviors effectively.

Enhanced prediction performance demonstrated.

Potential for detecting anomalous social network activity.

Abstract

The characterization and understanding of online social network behavior is of importance from both the points of view of fundamental research and realistic utilization. In this manuscript, we propose a stochastic differential equation to describe the online microblogging behavior. Our analysis is based on the microblog data collected from Sina Weibo which is one of the most popular microblogging platforms in China. Especially, we focus on the collective nature of the microblogging behavior reflecting itself in the analyzed data as the characters of the periodic pattern, the stochastic fluctuation around the baseline, and the extraordinary jumps. Compared with existing works, we use in our model time dependent parameters to facilitate the periodic feature of the microblogging behavior and incorporate a compound Poisson process to describe the extraordinary spikes in the Sina Weibo volume. Such distinct merits lead to significant improvement in the prediction performance, thus justifying the validity of our model. This work may offer potential application in the future detection of the anomalous behavior in online social network platforms.