Descriptive Answer Evaluation System using Natural Language Processing

In the education system, written exams play a vital role to test knowledge gained by students. But the total process from evaluating answers, grading, verification, and publishing results is time-consuming. A lot of human efforts are required for evaluating descriptive answers. This evaluation also depends on various factors like knowledge of the evaluator, application-level understanding of the evaluator. When any human being performs evaluation it varies along with the emotions of that person. So the goal of this paper is to develop the descriptive answer evaluation system which ensures uniformity in marking, saves a lot of time, and trouble of checking bundles of papers. In this evaluation system, first pre-processing techniques are applied to student’s answers. Then answers are matched against manually entered model answer using different techniques like keyword matching, Bert and finally score is calculated.