The dataset originates from the research results of T. Wambsganss, C. Niklaus, M. Söllner, S. Handschuh and J. M. Leimeister available here.
The dataset is originally in the brat standoff format and was converted to JSONL using pybrat.
During the training and evaluation of the data, it became apparent that the data was relatively heterogeneous and noisy. Due to this fact, the trained model only achieved an accuracy… See the full description on the dataset page:
https://huggingface.co/datasets/samirmsallem/argumentative_student_peer_reviews.
