Predicting Online Protest Participation of Social Media Users

Social media has emerged to be a popular platform for people to express their viewpoints on political protests like the Arab Spring. Millions of people use social media to communicate and mobilize their viewpoints on protests. Hence, it is a valuable tool for organizing social movements. However, the mechanisms by which protest affects the population is not known, making it difficult to estimate the number of protestors. In this paper, we are inspired by sociological theories of protest participation and propose a framework to predict from the user’s past status messages and interactions whether the next post of the user will be a declaration of protest. Drawing concepts from these theories, we model the interplay between the user’s status messages and messages interacting with him over time and predict whether the next post of the user will be a declaration of protest. We evaluate the framework using data from the social media platform Twitter on protests during the recent Nigerian elections and demonstrate that it can effectively predict whether the next post of a user is a declaration of protest. Social media has emerged as a popular information and communication channel for protest-related issues (Muthiah et al. 2015; Contractor et al. 2015; Tufekci and Wilson 2012). It provides an open and accessible platform for people to put forth views on issues affecting them. Millions of people, therefore, use social media to declare protest, mobilize opinion and participate in discussions on these issues. For example, electoral malpractice was suspected during the recent elections in Nigeria (Mark 2014), and users employed social media to express and mobilize viewpoints. Owing to this, social media can be used by protest organizers to recruit potential members. We concentrate on posting behavior of users and define online protest participation as the act of declaring protest through status messages. However, it is a nontrivial problem to directly predict online protest participation of social media users. Millions of users post in social media during popular social movements, and protest organizers have to go through them to identify potential participants. The mechanisms by which the protest affects the population cannot be fully observed (Lin et al. 2013), and protest organizers cannot easily estimate the number of protestors. Designing algorithms to predict whether the next post of a user will be a declaration of protest will enable protest organizers to anCopyright © 2016, Association for the Advancement of Artificial Intelligence (www.aaai.org). All rights reserved. ticipate the behavior of the protestors and estimate the total number of participants. This task faces several challenges. First, the ways in which protest-related events affect a person are not observable, resulting in a lack of knowledge of factors operating at that time causing his next post to be a declaration of protest. Second, a user is subject to various types of influence in his past, and many of them are in conflict with each other. This may lead to ambiguities on whether his posts will contain declarations of protest in the future. Finally, each user can post a large amount of content and interact with many people, leading to issues of scalability. Sociological studies have theorized factors from an individual’s history causing his next post to be a declaration of protest. A user will be more likely to protest if his social ties have reached out to him with protest-related messages in his past (Schussman and Soule 2005). The chances of him protesting are bolstered if these messages are sent by people interested in protest related issues (Lim 2008). The likelihood of protest is reduced if people uninterested in protestrelated issues have reached out to him in the past with messages unrelated to protests. (Snow, Zurcher Jr, and EklandOlson 1980). Inspired by these sociological theories, we utilize the user’s previous status messages and messages interacting with him to predict whether his next post will be a declaration of protest. To model the effect of the interactions on the user’s status messages, we build upon concepts from the Brownian motion theory of fluid particle motion (Zhou 2003). This theory models the path of fluid particles as other fluid particles come in contact with them over time. We draw analogies from these concepts to model the probability of the user declaring protest as other users reach out to him over time. The primary contributions of this work are: • Formally defining the problem of predicting whether the next post of a user will be a declaration of protest based on past status messages and messages interacting with him, • Demonstrating the applicability of sociological theories of protest participation in online social media data, • Proposing a framework that predicts online protest participation of a user, and • Evaluating the framework using a real world dataset from Twitter on the Nigerian election protest in March 2015. Figure 1: The proposed framework to predict if the next post of user u is a declaration of protest