Automated Essay Scoring (AES); A Semantic Analysis Inspired Machine Learning Approach: An automated essay scoring system using semantic analysis and machine learning is presented in this research

With the advancements in Artificial Intelligence (AI), ‘Automated Essay Scoring’ (AES) systems have become more and more prevalent in recent years. This research proposes an extension to the Coh-Metrix algorithm AES, with a focus on feature lists. Technical features, such as, referential cohesion, lexical diversity, and syntactic complexity are evaluated. Furthermore, it proposes the use of four novel semantic measures, including estimating the topic overlap between an essay and its brief. A prototype implementation, using neural networks, is used to test the individual and comparative performance of the newly proposed AES system. The results show a considerable improvement on the results obtained in the existing research for the original Coh-Metrix algorithm; from an adjacent accuracy of 91%, to an adjacent accuracy of 97.5% (and a QWK of 0.822). This suggests that the new features and the proposed system have the potential to improve essay grading and would be a good area for further research