Node Placement in Argument Maps: Modeling Unidirectional Relations in High &amp; Low-Resource Scenarios

Jundi, Iman; Falk, Neele; Vecchi, Eva Maria; Lapesa, Gabriella

doi:10.18653/v1/2023.acl-long.322

Cited by 1 publication

(1 citation statement)

References 0 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Many of them relate to quality aspects we consider in this work, from clarity and organization (Wachsmuth et al, 2016) to the general evaluability of arguments (Park and Cardie, 2018), potential fallacies in their reasoning (Goffredo et al, 2022), and the appropriateness of the language used (Ziegenbein et al, 2023). Recently, (Skitalinskaya and Wachsmuth, 2023) tackled the question whether an argumentative claim is in need of revision, whereas Jundi et al (2023) investigated where to best elaborate a discussion. While leverage claim generation for a refined assessment of argument quality, we are not aware of any prior work that actually optimizes arguments or their components in order to improve quality.…”

Section: Related Workmentioning

confidence: 99%

Claim Optimization in Computational Argumentation

Skitalinskaya,

Spliethöver,

Wachsmuth

2023

Proceedings of the 16th International Natural Language Generation Conference

View full text Add to dashboard Cite

An optimal delivery of arguments is key to persuasion in any debate, both for humans and for AI systems. This requires the use of clear and fluent claims relevant to the given debate. Prior work has studied the automatic assessment of argument quality extensively. Yet, no approach actually improves the quality so far. To fill this gap, this paper proposes the task of claim optimization: to rewrite argumentative claims in order to optimize their delivery. As multiple types of optimization are possible, we approach this task by first generating a diverse set of candidate claims using a large language model, such as BART, taking into account contextual information. Then, the best candidate is selected using various quality metrics. In automatic and human evaluation on an English-language corpus, our quality-based candidate selection outperforms several baselines, improving 60% of all claims (worsening 16% only). Follow-up analyses reveal that, beyond copy editing, our approach often specifies claims with details, whereas it adds less evidence than humans do. Moreover, its capabilities generalize well to other domains, such as instructional texts.

show abstract

Section: Related Workmentioning

confidence: 99%