Who is sharing the news in the Dutch Twittersphere? An exploratory analysis of news sharers’ self-presentation

Twitter is used as a platform for consuming and sharing news increasingly, since its launch in 2007 (Dagoula, 2020). While the platform's contribution to journalism can be argued to be both facilitating and destructive (Lee 2020; Broersma et al., 2010), it is obvious that Twitter is playing a significant role in the online news industry. A recent study by Reuters (2021) demonstrates that about 90 percent of the Dutch population relies on social media for their news information. From this group, about one third specifically relies on Twitter to educate themselves on current day issues and news (Digital News Report 2021). This demonstrates the significance of the platform in (online) news. Previous research has focused mainly on news engagement on Twitter (Chan et. al. 2021; Wischenewski 2021; Valenzuela et. al. 2017), and the motivation behind sharing news on social media platforms (Goh et. al. 2017; Kalsnes & Larsson 2017; Bhagat & Kim 2022).

However there is a lack of research on who specifically is sharing news on Twitter. News online is a reproduction of the gatewatching; individual bloggers and communities of commentators who may not report the news first-hand, but curate and evaluate the news and other information provided by official sources (Bruns, 2011). Since this process is critical in forming an individual's news diet (Bruns, 2018), investigating and understanding who is sharing news online is an important addition to understanding the role of twitter in news consumption and creating public opinion. We focus on users that are actively sharing news through hyperlinks on Twitter. We present a methodological approach exploring whether coherent groups of users that share similar news sources on Twitter can be identified. Gaining more insight into who is sharing news will provide a new basis to study the importance of identity and networks in creating public opinion online.

Our dataset consists of 10 million tweets from the year 2021, retrieved through the Twi-XL infrastructure. We use a mixed-method approach applying contextual topic modeling and clustering to group users based on their biographies, as well as manual explorations of the data in order to detect possible patterns in the categorization of the users and the news content that they share. By interpreting patterns in topics, clusters, and shared content we aim to unravel relationships between self presentation of news sharing Twitter users and news content that they share. Using Zero-ShotCTM and Agglomerative Clustering, we identified 15 topics and 6 clusters. We found that the topic containing the most links to social media platforms also contained the most references to news platforms. We also observe that, although with different proportions, each topic contains users from every cluster. We conclude that there is no clearcut and specific user profile representing "news sharers", and that many different people share news. However, politically involved users and news professionals are those that compose the best defined clusters. Future work will explore a combination of self-presentation and content tweeted to refine the identification of groups and their impact in the promotion of specific news topics.