ArgRewrite V.2: an annotated argumentative revisions corpus

Abstract

Analyzing how humans revise their writings is an interesting research question, not only from an educational perspective but also in terms of artificial intelligence. Better understanding of this process could facilitate many NLP applications, from intelligent tutoring systems to supportive and collaborative writing environments. Developing these applications, however, requires revision corpora, which are not widely available. In this work, we present ArgRewrite V.2, a corpus of annotated argumentative revisions, collected from two cycles of revisions to argumentative essays about self-driving cars. Annotations are provided at different levels of purpose granularity (coarse and fine) and scope (sentential and subsentential). In addition, the corpus includes the revision goal given to each writer, essay scores, annotation verification, pre-and post-study surveys collected from participants as meta-data. The variety of revision unit scope and purpose granularity levels in ArgRewrite, along with the inclusion of new types of meta-data, can make it a useful resource for research and applications that involve revision analysis. We demonstrate some potential applications of ArgRewrite V.2 in the development of automatic revision purpose predictors, as a training source and benchmark.

Predicting Desirable Revisions of Evidence and Reasoning in Argumentative Writing

“…Table 1: Example of revisions extracted from an essay from our elementary-school dataset. Kashefi et al, 2022) and understanding the pattern of revisions (Afrin and Litman, 2019;Shibani et al, 2018). Exploring the pattern of iterative revision have also been studied in scientific writing (Du et al, 2022).…”

To Revise or Not to Revise: Learning to Detect Improvable Claims for Argumentative Writing Support

“…The only works we are aware of that study revisions of argumentative texts are those of Afrin and Litman (2018), Skitalinskaya et al (2021), and Kashefi et al (2022). The first two suggest approaches that enable automatic assessments of whether a revision can be considered successful, that is, it improves the quality of the argumentative essay or claim.…”

SemEval-2022 Task 7: Identifying Plausible Clarifications of Implicit and Underspecified Phrases in Instructional Texts

“…Thus, the task follows a number of existing research contributions that deal with text revisions. A number of previous works examine reasons for and types of revisions in, for example, Wikipedia (Bronner and Monz, 2012;Daxenberger and Gurevych, 2012;Yang et al, 2017;Faruqui et al, 2018) and the essay-based corpus ArgRewrite Litman, 2015, 2016;Afrin and Litman, 2019;Kashefi et al, 2022). In this work, we use revision histories of instructional texts because clarifications seem particularly relevant in this domain.…”