Regional properties of global communication as reflected in aggregated Twitter data

Abstract

Twitter is a popular public conversation platform with world-wide audience and diverse forms of connections between users. In this paper we introduce the concept of aggregated regional Twitter networks in order to characterize communication between geopolitical regions. We present the study of a follower and a mention graph created from an extensive data set collected during the second half of the year of 2012. With a k-shell decomposition the global core-periphery structure is revealed and by means of a modified Regional-SIR model we also consider basic information spreading properties.adapting a simple information propagation model we can further differentiate the most efficient information sources of the world -as reflected in high volume of individual conversations. Methods Twitter dataWe used a data set collected from the freely available Twitter stream during the second half of the year of 2012 [3]. We refer to Twitter users in this data set allowing public access to their geographical location information as the geousers, and each one is located to a single fixed position [3]. Those that fell into unmapped territories (e.g., oceans) were discarded from further analysis. By means of aggregation of their different forms of connections we create two different communication networks between geopolitical regions: the mention (M) and the follower (F ) networks.For creating the user-level follower graph we used 177, 176, 790 links between 3, 312, 961 geo-users identified as the most active ones. Given their list, the additional follower relations were collected separately [3]. The source of a following link is the followed user while its target is the follower. A network built using the inverse follower relation is also be meaningful, and can be used for showing the direction of interest. To create the user-level mention graph we used 132, 436, 279 mention messages between 5, 381, 565 geo-users. The source of a mention link is the sender while its target is the mentioned user.

Spatially embedded real-world networks: Structure and dynamics from global to urban scales

“…By means of aggregation of their different forms of connections we created two different communication networks: the mention (M) and the follower (F ) networks Here we show two regional graphs based on [2]. We created the two graph representations, F and M, as weighted and directed graphs.…”

“…A comparative analysis of intra-regional and interregional communication ties can reveal structural properties, and be indicative of the information flow on the network. [2] We examine empirical centrality measures of nodes. First, we consider their generalized weighted degrees, or volume of directed communication ties as follows.…”

“…Similarly, the asymmetric mention centralities will measure if the region is dominantly a conversation starter or not. [2] When looking at the ratio of these connection numbers, where applicable (i.e., without division by zero), the asymmetry shows power law dependence of the number of internal connections. That means, that the greater the size of the region as measured by the regional self-links, the greater the connectedness to the other regions in both directions.…”

Inferring the interplay between network structure and market effects in Bitcoin

“…The growing availability of digital traces of human activity provide unprecedented opportunity for researchers to study complex phenomena in society [1][2][3][4][5][6]. Data mining methods that preform unsupervised extraction of important features in large datasets are especially appealing because they enable researchers to identify patterns without making a priori assumptions [6,7].…”

Propagation regularity of hot topics in Sina Weibo based on SIR model — A simulation research

“…Lerman et al[6] specifically extracted social networks of active users on Digg and Twitter, and tracked how interest in news stories spreads among them, they also showed that social networks play a crucial role in the spread of information on these sites, and that network structure affects dynamics of information flow. Kallus et al[7] introduced the concept of aggregated regional Twitter networksand characterized the communication between geopolitical regions. Liu et al [8] built a specific agent-based model using NetLogo based on the propagation model of SIR by taking consideration of the unique property of twitter-like websites in information flow, max spread time and the mechanism of people believing the rumor.…”