Built from 60 English scientific abstracts in the larger annotated dataset: Discourse Dependency TreeBank for Scientific Abstracts (SciDTB; Yang & Li, 2018), Accuosto and Saggion (2019) offered a fine-grained annotated dataset for the argumentative component classification and relation classification tasks, based on the study of Kirschner et al. (2015).
The dataset is token-based and segmented into elementary units that… See the full description on the dataset page: https://huggingface.co/datasets/DFKI-SLT/scidtb_argmin.
