Automated Grading System Using Natural Language Processing

Most of the articles which cover automated grading consider keyword matching to be a crucial aspect while grading answers. Even though these are important, it is human to forget several uncommon terms and instead replace them with words that have a similar meaning. In this paper, a solution to grading of papers of theory based subjects is obtained where in Automatic Paper Grading will be performed using Natural Language Processing. Machine learning techniques like Semantic Analysis will be adopted. As a single answer can be presented in a number of ways by different students, matching keywords is inefficient. That is why, using ontology, extraction of words and their synonyms related to the domain is done which makes the evaluation process holistic as presence of keywords, synonyms, the right word combination and coverage of concepts can now be checked. The above mentioned techniques will be implemented with Ontology and will be tested on common input data consisting of technical answers. The results will be analyzed and an unbiased, high accuracy automated grading system for a theory based subject will be obtained with very little error rate which is comparable to a differential human-to-human error rate. The algorithm is designed based on the responses collected during the survey conducted amongst teachers regarding their parameters when correcting papers manually.