In the digital era that we live in, online discussion has become more and more present in our daily lives and especially on the social media platforms that we frequently use. Whenever we post an image on Instagram or write a twit on Twitter, we start a potential discussion thread. How come is that?
When we publish a message (a post, an image…) and share it with other people, if someone makes a comment then we generate a conversation thread. Nothing new until here, right? What’s new and interesting (for me) is that this sequential posting behaviour has attracted the attention of academia (besides the big tech companies, of course).
It seems that different modelling approaches have been proposed to study the structure of threads in online discussions in order to identify the mechanisms underneath the network structure of threads. In particular, statistical models are the most popular and allow scientists to reproduce the growth of discussion threats by using different features/parameters and link them to human behaviour. The parameters of these models can help to compare different social media platforms and assess the impact of design choices and user interface changes on the way the discussions unfold. Also, the models can provide interesting insights into human behaviour and spread some light into the relevant social theories explain behaviour in online discussions, such as homophily, emotional contagion, or social influence. In other words, these particular models can help to better understand the current theories on the table.
For example, the image below is from a reply network of users on Wikipedia article talk pages from Iosub et al. (2014). The colour of nodes expresses the proportion of words expressing anger (from blue to red). Assortativity observed in this network (e.g. clusters of red nodes) might be explained by either homophily or emotional contagion.
Featured image from Pexels – CC0.