Purpose: The purpose of this paper is to describe and analyse relationships and communication between Twitter actors in Swedish political conversations. More specifically, the paper aims to identify the most prominent actors, among these actors identify the sub-groups of actors with similar political affiliations, and describe and analyse the relationships and communication between these sub-groups.

Design/methodology/approach: Data were collected during four weeks in September 2012, using Twitter API. The material included 77,436 tweets from 10,294 Twitter actors containing the hashtag #svpol. In total, 916 prominent actors were identified and categorised according to the main political blocks, using information from their profiles. Social network analysis was utilised to map the relationships and the communication between these actors.

Findings: There was a marked dominance of the three main political blocks among the 916 most prominent actors: left block, centre-right block, and right-wing block. The results from the social network analysis suggest that while polarisation exists in both followership and re-tweet networks, actors follow and re-tweet actors from other groups. The mention network did not show any signs of polarisation. The blocks differed from each other with the right-wingers being tighter and far more active, but also more distant from the others in the followership network.

Originality/value: While a few papers have studied political polarisation on Twitter, this is the first to study the phenomenon using followership data, mention data, and re-tweet data.

The aim of this article is to engage with problems of sampling and completeness currently discussed within data science through the specific example of conversations in Twitter. Some of the difficulties involved in Twitter concern restrictions laid out by platform owners, restrictions that make it difficult for researchers to collect complete conversations. A contribution is made through the development of a method for collecting and analyzing follow-on conversations around a set of hashtags. This was made possible through the simultaneous tracking of a set of hashtags and prominent participants in the conversation. The full set of tweets was compared to the subset of tweets including either of the selected hashtags. Including follow-on conversation increased the set of tweets by 56% and the set of tweeting users by 17%. It is also shown that different network analysis techniques and filtering options give different results with regard to prominent users.

Introduction. Following the news of an extraordinary agreement between six political parties, intense discussions on Twitter took place, suitable for investigation of political controversies on the platform. The purpose is to analyse threaded conversations on Twitter in relation to this political event.

Method. By using the streaming API with a set of hashtags and the most active participants in the conversations, tweets related to the event and tweets belonging to the follow-on conversation were captured. From this set, the replies were used to build conversational threads.

Analysis. The dataset was analysed using both quantitative and qualitative methods. The overall lifecycle of the conversations was outlined using statistical methods. Ten conversational threads were studied using qualitative content analysis.

Results. Immediately after the agreement, activity was focused on information diffusion, but following this, discussions emerged. Politicians were frequently talked to but rarely replied to tweets directed to them. Citizens from the general public dominated the activity.

Conclusions. Conversations do exist beyond the hashtag but few examples of democratic debate were found. Conversations are more likely to develop within tightly knit groups followership-wise. Echo chambers as well as discussions where non like-minded argued were identified. Consensus formation was not common.

This is a cross-field literature review and comparison of the fields webometrics (cybermetrics) and web (data) mining.