The paper, ‘A Computational Analysis of Vagueness in Revisions of Instructional Texts,’ explores the concept of vagueness in instructional texts and how revisions can address this issue. The researchers use the wikiHowToImprove dataset, which includes revision histories of instructional articles, to investigate changes in vagueness before and after a revision. They also apply a neural model to distinguish between the original and revised versions of an instruction. The study contributes to the creation of a dataset of vague and clarified instructions, provides an analysis based on semantic frames, and demonstrates the initial results of a neural model’s ability to differentiate between the two versions.
Publication date: 21 Sep 2023
Project Page: https://arxiv.org/abs/2309.12107v1
Paper: https://arxiv.org/pdf/2309.12107
