Social network of extreme tweeters

Abstract

The number of posts made by a single user account on a social media platform Twitter in any given time interval is usually quite low. However, there is a subset of users whose volume of posts is much higher than the median. In this paper, we investigate the content diversity and the social neighborhood of these extreme users and others. We define a metric called "interest narrowness", and identify that a subset of extreme users, termed anomalous users, write posts with very low topic diversity, including posts with no text content. Using a few interaction patterns we show that anomalous groups have the strongest withingroup interactions, compared to their interaction with others. Further, they exhibit different information sharing behaviors with other anomalous users compared to non-anomalous extreme tweeters.

A Query-Driven System for Discovering Interesting Subgraphs in Social Media

“…The three rows in Figure 10 show the network characteristics of the subregions of this graph, related, respectively to the themes of #ados and "black lives matter" (Row 1), Captiol insurrection (Row 2) and Socioeconomic issues related to COVID-19 including stimulus funding ad business reopening (Row 3). In earlier work (Zheng and Gupta, 2019), we have shown that a well known propagation mechanism for tweets is to add user-mentions to improve the reach of a message -hence the user-mention subgraph is indicative of propagative activity. In Figure 10, we compare the hashtag activity (measured by the Hashtag subgraph) and the mention activity (size of the mention graph) in these three subgraphs.…”

Discovering Interesting Subgraphs in Social Media Networks

“…For S i to be interesting, it must have at least one attribute a such that D(C(a)) does not have the usual power-law distribution expected in social networks. Zheng et al [2] presents two such measures over tweet text -vocabulary diversity (distribution of distinct nonstop-word terms) and topic diversity (computed as SVD vectors). They showed that interesting tweets show a significantly low diversity compared to those of "standard" tweet collections.…”

“…Once these candidate graphs are constructed, they are tested for criterion C3. In this paper, we have directly applied the diversity metric proposed in [2].…”

“…Even if there are similarly dense subgraphs representing an intense discussion on the theme, it is not necessarily interesting, unless the content of the conversation is very different from the content of the conversation of the graph surrounding it, which would be the case if everyone else in a network discusses politics while this group discusses a rapper or if the content of this group is extremely narrow. A version of this problem was investigated in [2], focused on "extreme tweeters", users who tweet a lot, have a very close but highly active network, but have a very limited discussion vocabulary. Thus, the notion of interestingness depends both on the content and structure of the subgraph and can only be interpreted in the context of a reference network as determined by an analyst's need.…”