Trends of news diffusion in social media based on crowd phenomena

Abstract

Information spreads across social media, bringing heterogeneous social networks interconnected and diffusion patterns varied in different topics of information. Studying such cross-population diffusion in various context helps us understand trends of information diffusion in a more accurate and consistent way. In this study, we focus on realworld news diffusion across online social systems such as mainstream news (News), social networking sites (SNS), and blogs (Blog), and we analyze behavioral patterns of the systems in terms of activity, reactivity, and heterogeneity. We found that News is the most active, SNS is the most reactive, and Blog is the most persistent, which governs time-evolving heterogeneity of these systems. Finally, we interpret the discovered crowd phenomena from various angles using our previous model-free and model-driven approaches, showing that the strength and directionality of influence reflect the behavioral patterns of the systems in news diffusion.

News Sharing in Social Media: A Review of Current Research on News Sharing Users, Content, and Networks

“…Research on news sharing networks is highly focused on technological aspects and thus dominated by scholars from the computer and information sciences. Hence, the analyzed studies try, for example, to uncover topological or temporal characteristics of news sharing (e.g., Kwak, Lee, Park, & Moon, 2010; Peng & Marculescu, 2013), develop models or algorithms to predict news sharing cascades in social media (e.g., Goyal, Bonchi, & Lakshmanan, 2010; Lerman & Galstyan, 2008; Myers, Zhu, & Leskovec, 2012), or investigate how independent decisions by social media users ultimately lead to distinct network structures and sharing characteristics (e.g., Hu et al, 2012; Kim, Newth, & Christen, 2014). In this area of research, the term “network” is used to refer to two related but nonetheless distinct phenomena.…”

A Model for Scale-Free Networks: Application to Twitter

“…It can be further used as a starting point in simulating the information dynamics of the social network. There are many information-theoretic models that simulate the diffusion of messages [11][12][13][14][15][16]. They characterize the dynamics of retweeting activity, leading to interesting insights, such as users' tweeting patterns in relation to their degree distribution or the ability to predict whether a piece of information will be virally transmitted through the network.…”

Real-world diffusion dynamics based on point process approaches: a review

“…Accordingly, an individual item has been the focus for modelling its popularity dynamics, such as a hyperlink (65), a single tweet (39,40), memes (short textual phrases) (41), and an individual article in science (42,43). In addition, a collection of items have been also considered to predict the popularity of relevant items as a super set (53,70,71,89). For instance, multiple hyperlinks are assigned a single topic by recognizing named entities (e.g., person, organization, place) in the main content of the reference pages, resolving entities with data matching techniques (117), and classifying documents into relevant topics (5).…”