What once had to be laboriously coded by hand in content analyses, Gregor Wiedemann does in a fraction of the time. Together with his team, he is developing procedures that filter complex semantic structures in the form of arguments out of huge bodies of text.
This technique is called argument mining. It is helpful when you want to pick out very specific parts of large amounts of text, as we encounter online, in order to carry out a further qualitative analysis, for example.
Gregor Wiedemann has already used this method to examine the debate on nuclear energy in the online articles of the British Guardian. Currently, he and his team are extracting arguments from online debates on minimum wages and the legalisation of marijuana.
With ongoing digitalisation, automated content analyses such as argument mining are becoming increasingly necessary, because new texts are constantly being created in the digital space. Not only on social media or on online news portals: many print products are digitising their archives, making them accessible for automated content analysis. Thus, the amount of texts that can be researched is becoming incredibly large.
Gregor Wiedemann does not believe that automated text analysis will ever completely replace human researchers. There will always be a mixture of analogue and automated media research. He does not dare to say in which direction automated analysis procedures will develop. "Five years ago, I could not imagine that what is actually possible today would be possible. That is why I hardly dare to make a prognosis." The development is rapid. It is exciting to see where the road will take us in the future.
HBI project on argument mining
A Framework for Argument Mining and Evaluation (FAME)