Marcus Maurer / Johannes Daxenberger / Matthias Orlikowski / Iryna Gurevych

Argument Mining: A new method for automated text analysis and its application in communication science

In the age of online communication, automated methods for analyzing media content become more and more important in the social sciences. However, methods for automated content analysis used in communication science thus far are still limited to a small set of highly standardized categories like the central topic or the tone of a message and, therefore, fail to reach the complexity of manual codebooks. In this paper, we introduce argument mining, a new method for automated text analysis recently developed in computer science. Based on machine learning algorithms, Argument Mining detects full arguments in media messages. Moreover, the quality of these arguments can be measured. We give a brief introduction in the method and its underlying assumptions, discuss recent research in computer science based on Argument Mining, give an overview about various possible future applications of Argument Mining in communication science, and mention some limitations concerning the transfer of Argument Mining techniques to communication science issues.

Keywords: Automated Content Analysis, Argument Mining, Machine Learning, Argument Quality