Automated Claim Detection in Argumentative Essays and their Relationship with Writing Quality

EasyChair Preprint no. 3850

Abstract

This study extracted content and structural features to predict human annotations for claims and non-claims in argumentative essays. The evaluation of classification models indicated Gaussian Naive Bayes classifier yielded the most balanced identifications of claims and non-claims. We used the model to make predictions in a validation corpus that included human ratings of writing quality. The number of claims, the percentage of non-claims, and the average position of non-claims were significant indicators of essay quality.

Keyphrases: argument mining, claim detection, essay quality

@Booklet{EasyChair:3850,
  author = {Qian Wan and Scott Crossley and Laura Allen and Danielle McNamara},
  title = {Automated Claim Detection in Argumentative Essays and their Relationship with Writing Quality},
  howpublished = {EasyChair Preprint no. 3850},

  year = {EasyChair, 2020}}