Argumentative Relation Classification for Argumentative Dialogue Systems

This is my Bachelor's thesis with the respective implementations.
Every folder provides a README with further information.
Please cite my thesis when you use any of the material:

@mastersthesis{Schindler_ArgumentativeRelationClassification,
	type = {Bachelor's thesis},
	title = {Argumentative Relation Classification for Argumentative Dialogue Systems},
	author = {Carolin Schindler},
	year = {2020},
	school = {Institute of Communications Engineering, University of Ulm},
	doi = {10.18725/OPARU-33911}
}

Annotation

We executed an annotation study in order to evaluate the Argumentative Relation Classification Pipeline.

Argumentative Relation Classification Pipeline

Main contibution of this thesis.
Our pipeline covers the whole process from querying the argument search engine over argumentative relation classification to the output of the argumentation structure.

Computation Time

We investigeated the computation time for different approaches.

EVA Sample Discussion

We created a sample owl file with our argumentative relation classification pipeline and from this a sample discussion with the argumentative dialogue system EVA desigend by Niklas Rach et al.

Relation Model

We trained and evaluated different models to predict whether there is a relation from one sentence/argument to another.