What Sticks With Whom? Twitter Follower-Followee Networks and News Classification

In this paper we analyze Twitter as a news channel in which the network of followers and followees significantly corresponds with the message content. We classified our data into twelve topics analogous to traditional newspaper sections and investigated whether the spread of information depended upon the Twitter network of followers and followees. To test this, we mapped the social network related to each topic and calculated the occurrence of retweet and mention messages whose senders and receivers were interconnected as followers and followees. We found that on average 10% of retweets (RT-messages) and 5% of direct mentions between users (AT-messages) in Twitter hashtags are sent and received by users interconnected as followers and followees. These figures vary considerably from topic to topic, ranging from 15%-19% within Technology, Special Events and Politics to 3%-5% within the categories Personalities and Twitter-Idioms. The results show that hard-news messages are retweeted by a considerably larger community of users interconnected as followers and followees. We then performed a statistical correlation analysis of the dataset to validate the classification of hashtag in news sections based on retweet connectivity. 1. Twitter as a Source of News Recent literature has examined a number of approaches to information diffusion in Twitter. Previous studies (Bakshy, Hofman, Mason, & Watts, 2011; Huberman, Romero, & Wu, 2009; Kwak, Lee, Park, & Moon, 2010) have shown that Twitter’s topological features comprise a highly skewed distribution of followers and low rate of reciprocated ties. Influence on Twitter was found to be connected to network topology, even though metrics such as the number of followers, page-rank, and number of retweets presented different results (Kwak, et al., 2010; Wu, Hofman, Mason, & Watts, 2011). Copyright © 2012, Association for the Advancement of Artificial Intelligence (www.aaai.org). All rights reserved. This work was supported by São Paulo Research Foundation (Grant 2010/06243-4). Bakshy et al. (2011) investigated the distribution of retweet cascades on Twitter and determined that although users with large follower counts and past success in triggering cascades were on average more likely to trigger large cascades in the future, these features were in general poor predictors of future cascade size. Wu et al. (2011, p. 3) found that Twitter does not conform to the usual characteristics of social networks, which exhibit much higher reciprocity and far less-skewed degree distributions, but instead better resembles a mixture of mass communication and face-to-face communication. Kwak et al. (2010) crawled the entire Twitter network and found a non-power-law follower distribution, a short effective diameter, and low reciprocity, which all mark a deviation from the characteristics of human social networks described by Newman (2003). Kwak et al. also found that Twitter and Korean social network Cyworld present a much higher power-law distribution than most social networks. The characteristics shared by Twitter and Cyworld are that many celebrities are present and that they interact with their fan base. This characteristic emphasizes the importance of celebrities and media-pundit users in social networks such as Twitter. Kwak et al. (2010) encountered a short average path length that might be a symptom of Twitter’s role as an information mechanism, as users follow users not for social networking, but for information. The investigation of Wu et al. (2011) was consistent with the results of Kwak et al. (2010) regarding the topological features of Twitter followers graph. They concluded from the highly skewed nature of the distribution of followers and the low rate of reciprocated ties that Twitter more closely resembled an information sharing network than a social network. The question of whether Twitter better resembles an information sharing network or a social network was also addressed by exploring the variety of topics that flow throughout the Twitter network. Romero et al. (2011) examined the hypothesis that hashtags for different topics AAAI Technical Report WS-12-01 The Potential of Social Media Tools and Data for Journalists in the News Media Industry