News websites often have rich social interaction opportunities, e.g., comment sections on their websites. Do readers agree? What arguments are put forward, and what significance do they have in the discourse? Do some statements polarize more than others? Semantic analysis helps to comment analysis in a time-saving and meaningful way.
Case Study "Homeopathy" Discourse on 20 Minutes
In December 2024, the media platform 20 Minutes reported on viewer reactions to a PULS broadcast by the Swiss broadcaster SRF on homeopathy. The broadcast sparked a broad debate. Should homeopathy remain a health insurance benefit, or is it just "quackery"? The in-house comment section filled up within a very short time with posts that triggered a large number of social reactions. We have analyzed this case as an example and show how comments can be meaningfully evaluated.
Metrics for comments
Within a very short time, over 1044 readers rated the criticism of the program's statements. 421 comments were posted and led to discussion after the program. The article was forwarded a total of 219 times.
But what was the content of the debate and can we learn from it?
How hot and engaging is your topic of interest? The first indicator is the frequency with which posts are made. We determine the frequency to get an initial statement about the activation potential of a topic. Depending on the time of day an article is posted online, the first evaluable reaction pattern appears here. Most reactions occur in the first few hours and then level off. In addition, media such as 20 Minutes in Switzerland often turn off the comment function after just one day.
Some commentators post but get no response, intentionally or not. We call that broadcasting. The post is a one-way channel. A "conversation" in response to a comment never really started. In a unidirectional conversation, there is at least one response to the comment, and in bidirectional communication, there is even at least a brief exchange between readers. All three types of communication are the basis for us to conduct an entity-relationship analysis, which visually displays interaction patterns. People are nodes, and responses are edges that connect them. The metrics that measure the proportion of our communication types characterize the overall conversation and the level of interaction and offer great insights into the activation potential of the subject in question.
Comments can polarize
When analyzing comments, we use the available data from the comment systems of a media outlet in focus. 20 Minutes, for example, allows the evaluation of comments in a six-level category system, where three are positive and three are negative. From this interaction option, we calculate the polarization score.
Articles can generally be met with almost complete approval (polarization values up to "1") or provoke complete rejection or disapproval (polarization values up to "-1"). In the middle, with values around "0", the approval and rejection of a statement cancel each other out. This is where the online debate splits into two camps. These are statements that most often lead to the formation of camps and are, therefore, very interesting from a communicative point of view.
We typically list the articles with the highest social reaction in the evaluation according to their polarization. This helps us better understand the form of discourse on important topics and the argumentation structures. Where do readers agree? Where do they reject an argument? And where are they divided and have diverging opinions?
Summary Comment Analysis
With comment analysis, we help you to better understand how the market deals with your relevant topics. We have the technology to evaluate relevant media sites on the Internet, with the help of which we can provide you with the relevant decision-making basis from a communication perspective so that you can better plan your communication strategy.
We can use metrics to provide fact-based knowledge so that you don’t fall into communication traps.
Interested in learning more?