Abstract

Motivated by the mandate to design and deploy a practical,real-world educational tool for grading, we extensively ex-plore linguistic patterns for Short Answer Scoring (SAS) aswell as authorship feedback. We approach the SAS task viaa multipronged approach that employs linguistic context fea-tures for capturing domain-specific knowledge while empha-sizing on domain agnostic grading and detailed feedback viaan ensemble of explainable statistical models. Our method-ology quantitatively supersedes multiple automatic short an-swer scoring systems.