To the issue of corpora methods in automating argumentation analysis using the example of scientific peer review
Abstract:
As part of the study of argumentation in scientific peer review, we formalized the process of finding solutions to disputes between authors and reviewers using logical inference based on abstract argumentation frameworks with trust-based extension semantics. The proposed method was tested on the basis of specially annotated scientific peer reviews. The markup scheme for scientific peer reviews is based on a standard review template and an algorithm for generating their OWL representations, which together provide the formalization of argumentative disputes during scientific peer review using abstract argumentation frameworks and finding solutions to such disputes using automatic logical inference. The study revealed problems related to both the peculiarities of manual argumentative annotation of texts and the lack of annotated and labeled text corpora of scientific peer reviews in Russian. We propose to continue the research, taking into account the identified limitations, on an interdisciplinary basis with the participation of computer linguists.
