Machine Learning Driven Argument Mining
MCML Authors
Michael Fromm
Dr.
* Former Member
Abstract
Michael Fromm
Dr.
* Former Member
Abstract
This thesis addresses the challenges of argumentation in the digital age by applying machine learning methods to automatically identify, retrieve, and evaluate arguments from diverse and often contradictory online sources. The first focus is on argument identification, specifically in heterogeneous text sources and peer reviews, where the relationship between the topic and arguments is crucial, and knowledge transfer across domains is limited. The second focus is on argument retrieval, where machine learning is used to select relevant documents, ensuring comprehensive and non-redundant argument coverage. Finally, the thesis explores the strength or quality of arguments, integrating this concept with other argument mining tasks and evaluating its impact across different text domains and contexts. (Shortened.)
BibTeXKey: Fro22