This machine learning project contains the latest code version for the Argument Mining tasks researched by Matthias Liebeck as part of his PhD thesis.

The code of this project is archived to reflect the state of the PhD submission. Further development and refactoring can be found here.

• docker
• Execute the data/download_resources.sh script to download the models

./data/download_resources.sh data

• Depending on whether you want to use our trained Wikipedia word2vec models, you also need to download word embeddings via this script:

./data/download_embeddings.sh data

• If you want to use character embeddings from fasttext, you can download the specific model here: http://lager.cs.uni-duesseldorf.de/NLP/fasttext/german/wikipedia/de-wiki_20170501/
  With the following script, you can download chracter (3, 3)-grams that were trained with 5 and with 50 iterations:

./data/download_character_embeddings.sh data