The research paper delves into the area of counter-argument generation, a captivating area in computational linguistics. The primary focus is on sentence-level counter-argument generation, which offers unique constraints and challenges. The paper presents the ArgTersely benchmark for this task, using a manually annotated dataset from the ChangeMyView debate forum. A BERT-based evaluator Arg-Judge was trained for better evaluation, and a comparative experiment was conducted involving various baselines. The results show the competitiveness of the proposed framework and evaluator in counter-argument generation tasks.
Publication date: 22 Dec 2023
Project Page: https://github.com/amazingljy1206/ArgTersely
Paper: https://arxiv.org/pdf/2312.13608
