Online discussions and social networks on the internet have begun to play an important part in the political life of contemporary societies. This has made researchers interested in the field of network analysis. However, commenting in the blogosphere and co-commenting communities are areas which so far have been insufficiently investigated.
Olessia Koltsova, Sergey Koltsov and Sergey Nikolenko supposed that posts with similar subject matter would attract the same commenters. If someone is interested, for example, in recipes, they might comment on all or many posts about cooking, and other users would behave in the same way. So, around each post on recipes a cloud of comments from the same group of people would take shape. But if the cloud grows around an author rather than a topic then we can say that these authors are opinion leaders for a group of commenter-admirers.
To test these suppositions, the researchers looked at materials on Live Journal. They chose Live Journal in particular because in Russia a large part of the current socio-political discussions are concentrated here.
The selection to be analysed contained posts of 2,000 popular Live Journal bloggers over the period of a week. In the range of blog posts the researchers said they found there were enough comments to form ‘clouds’ or communities. “The experiments with ratings showed that already in 50,000th place, the number of comments was almost two times less than the number of posts and was fewer than 3 comments per blogger. If you take into account that these bloggers are not famous, the chances of commenting on different posts by the same person are extremely low”.
All previous research at the laboratory shows that the top 2,000 bloggers (about 1% of all accounts) average almost twenty times more comments than posts and in such sample the researchers noted that there are active commenters who leave a lot of comments.
The researchers chose a period of a week for analysis (1-7 April 2013) because, as they explained, most comments on a post are usually made within a few days. Once posts with no comments had been removed, the researchers focused their attention on 17,386 posts by 1,667 authors. There were 520,000 comments on them from about 56,000 bloggers.
The researchers applied methods of network analysis. Posts were represented on a graph by dots and lines between the dots indicated that two posts had comments from the same person. Thus, mathematical graphs were formed. Those parts of the graph where there are a lot of lines linking posts show a large number of the same commenters on more or less the same posts. A special community detection algorithm was used to highlight the parts on a big graph where there were tens of thousands of dots and hundreds of thousands of connecting lines. The term ‘community’ here means the most densely filled areas on the graph.
The analysis showed a large number (85) of small communities – dyads and triads, often isolated and not being of significant interest. The authors explain that this is one consequence of the uneven allocation of the number of comments per post and particularly the number of comments per commenter. About one third of commenters in the sample left only one comment and so don’t generally take part in the graph of co-commenting. The other group is made up largely of less than a thousand active commenters who generate from 100 to 1,000 comments each. This groups accounts for two thirds of all the comments. The largest community, like the smallest is not so interesting. It is a ‘free-for-all discussion about everything’. The seventy-odd medium-sized communities are potentially more interesting.
The research shows that a community has an expressed tendency to gather around authors of posts which are commented on, who can therefore be considered opinion leaders. ‘It’s true, their local, applied to the given community leadership could have a latent character, as the community itself might not realise who its members are,’say the researchers. This is possible because by commenting on various posts by one author, users might not engage with other bloggers or, at least, might not be aware of the limits of the circle of people who actively comment on that author.
The subject matter of commented posts draws a community around it to a lesser extent. The researchers think that communities united around a particular issue seem to arise when the authors of posts or group of authors of posts, who predominate in a community, write consistently on one theme. But confirmation or rejection of these connections will be the first question for further research.
In any case, the results of this analysis show that discussions on specific issues, although they do exist are clearly not predominant. Now the researchers a faced with determining what is the relationship between themed and non-themed discussions and are there sufficient numbers of themed discussions to be socially significant and worthy of study. Besides if the share of these communities is sufficient, what is the best way to distinguish them from the noisy ‘free-for-alls’?