Using YouTube Data to Display Comment Networks among Organisations Associated with Protest Events

In this short post, we are introducing another type of resource that The TakePart team has been producing, namely topic-based social media network graphs. We are using a well-established approach in social movement studies, Protest Event Analysis (PEA), to map out protests that have taken place in the five project countries (Germany, Hungary, Romania, Spain and the UK) from January 2023 to the end of July 2024 (the study period). Within this period, we catalogued protests reported by media outlets in the project countries, from across the ideological spectrum. Following that, we relied on a set of bespoke instruments to retrieve the social media footprint of the organisations reported by the media as being involved in protests, during the study period. While this process is ongoing, we are collecting social media data that we hope will enable us to distinguish how these organisations and the people who engage with their content, on social media, communicate about protest in contrast to other types of content. 

In this post, we present an initial analysis of comments associated with the most prominent protest-related topic (i.e., that contained the word ‘protest’) in the content produced by organisations in our YouTube dataset for Romania (see Figure 1 for the distribution of these topics); and second, the most prominent topic among comments on all the content produced by the same organisations (see Figure 2a for the distribution of these topics). To do this, we used the machine-learning BERTopic technique for modelling latent topics in a collection of texts. We thus produced a hierarchy of topics based on the frequency of their occurrence. We then visualised the most prominent of them, namely the opposition to the ‘special [read handsome] pensions’ that judges and other professional groups employed in the public sector receive (see for example this EU Commission briefing paper  and Figure 2b for the most frequent 10 terms associated with the ‘special pensions’ topic). An example of a comment posted on YouTube that fits under this topic was, ‘’TAIATI PENSILE JEGOASE LA JAVRE SA VA FIE RUSINE HOTILOR’ (‘cut the dirty pensions of those scoundrels. You should be ashamed of yourselves!’, original emphasis).

 

 

Figure 2a. Frequency count for the top 50 topics among comments in the organisational YouTube dataset for Romania [Topic 29 (most frequent topic): ‘Special Pensions’)]. 

Figure 2a. Frequency count for the top 50 topics among comments in the organisational YouTube dataset for Romania [Topic 29 (most frequent topic): ‘Special Pensions’)]. 

Figure 2b. Topic 29 (all comments) and Topic 36 (protest-related topic): Key Terms in the Most frequent Topics

Following that, we visualised the networks of comments to video posts about the special pensions by the organisations in our dataset. The organisations span the spectrum from media to political party and civil society organisations. The visualisations in Figures 3 and 4 show that people who commented on posts produced by these organisations helped link them into a large, connected, network component.

The network visualisation in Figure 3 provides an insight into, first of all, the key actors involved in this protest-related communication. One thing that immediately stands out is that the most important nodes in the large network component in that figure are an established media organisation (‘Romania TV’) and a political party, USR, that has historical links to social movements, in Romania (see for instance Mercea & Santos, 2024). Other media organisations and parties are likewise central to the network: ‘Realitatea TV’ as well as ‘Libertatea’ (a national daily) along with local news outlets such as ‘Ziarul de Iasi’, ‘Ziarul de Cluj’ and ‘Ziarul Unirea’; and, on the other hand, parties such as ‘Partidul REPER’ (a recently formed centre-right party), ‘Partidul Social Democrat’ (centre left and the main partner in the governing coalition, in 2023-24), ‘Alianta pentru Unitatea Romanilor’ (AUR, a far-right party) as well as party leaders such Dominic Fritz, the USR mayor of Timisoara. Lastly, civil society organisations (CSOs) such as ‘Agent Green’ (an environmental NGO), ‘Declic’ (an online petition platform) and ‘Fratia Ortodoxa Sfantul Mare Mucenic Georghe’ (a far-right NGO) are less central but still part of this connected component.

The visualisation in Figure 4 represents all the comments on all the posts produced by the organisation in our study, clustered under the most prominent topic (and not only those explicitly related to protest). A quick glance at the figure reveals a very similar connected component to that in Figure 3 (where the most central nodes are Romania TV and USR). This is not a coincidence in as far as the most prominent topic among all comments was also the special pensions. The following quotes is an illustration of the comments clustered under this topic:

‘Daca esti asa tare cu pensiile speciale de ce nu ai facut nimic?’, [If you are the honcho that you say you are then why didn’t you do anything about the special pensions?]

 ‘Fără  pensii  speciale   !!!’, [‘Get rid of the special pensions’]

 ‘Din cauza lor nu merge nimic in tara ,,,nu mai au bani de pensii speciale’ [‘It’s thanks to them that nothing works in this country,,, they ran out of money for special pensions’]. 

While the comments speak to a sense of anger and aggravation regarding the special pensions, the way in which they connect these different organisations is notable in at least a couple of respects. First, we see that the topic itself stirred up a lot of reactions, being the most commented among all the posts by the organisations in our dataset. Second, those organisations encompassed a large diversity of not only CSOs but also mainstream media and political parties, the latter of which also spanned much of the ideological spectrum. Not having yet analysed the content of those comments in any depth, we can only intimate that these organisations, while connected, sat on opposing sides of the debate on special pensions. However, the way in which they were implicated in that debate may not be all that straightforward. For example, in Figure 3, while many people who commented on posts by USR also commented on posts by Romania TV, people who commented on posts by Fratia Ortodoxa engaged more with the content produced by Romania TV than by USR. Conversely, people who commented on posts by Declic and Agent Green engaged more with content by USR, Dominc Fritz, Partidul REPER as well as Realitatea.Net. This placement points to some degree of ideological clustering within this connected component.

Going forward, we will continue to explore these results and what they mean for the analysis of communication surrounding protest events. Our focus will turn to investigating social learning in such networks and any implications for civic participation.  

Figure 3. Visualisation of the YouTube comment network related to protests on special pensions by organisations in Romania involved in protest events from January 2023 to July 24. 

Figure 4. Visualisation of the YouTube comment network formed around posts regarding special pensions by organisations in Romania involved in protests from January 2023 to July 24.