Automatic Short -Answer Grading System (ASAGS)

TL;DR

This paper presents an automatic short-answer grading system that leverages semantic similarity techniques, demonstrating significant improvements over traditional lexical matching methods in accuracy.

Contribution

The paper introduces a semantic-based approach for automatic short-answer grading that outperforms traditional lexical similarity methods.

Findings

Semantic ASAGS achieves up to 59% improvement over vector-based similarity.

Experiments validate the effectiveness of semantic techniques in short-answer assessment.

The system outperforms simple lexical matching methods in accuracy.

Abstract

Automatic assessment needs short answer based evaluation and automated assessment. Various techniques used are Ontology, Semantic similarity matching and Statistical methods. An automatic short answer assessment system is attempted in this paper. Through experiments performed on a data set, we show that the semantic ASAGS outperforms methods based on simple lexical matching; resulting is up to 59 percent with respect to the traditional vector-based similarity metric.