Automatic Evaluation of Descriptive Answer using Optimized Neural Network Algorithm

As everything in the modern world is automated, examination processes also need to be automated. The Automatic Evaluation system is already established for objective questions, where students can select any choice as the answer and mark it. Since the answer is simple to record and the system evaluates and assigns a grade in an effective manner, the automatic evaluations of this are similarly quite simple. However, descriptive answers are required in order to fully comprehend the student's depth of knowledge, where students can write about their in-depth understanding of the solutions. The manual evaluation of objective type of answers is also very simple because the answer is a single word and the evaluator easily correct the papers and put the marks. Descriptive answer evaluation requires more time and requires additional staff members. Sometimes students will write multiple answers to the same question depending on how students desire to express themselves, and the human evaluators may have fluctuating emotions when assessing the papers, which leads to erroneous evaluations of descriptive answers. In order to address this, we have created a model that automatically evaluates answers that are descriptive in nature and assigns a score between 0 to 10. For this motive, a model is created that accurately assesses students' exam performance through the use of natural language processing and an inventive artificial neural network technique. Similar to a human evaluator, this model is designed for an assessment system and produces results that are more accurate and consistent. When this model is compared to the evaluation scores of human evaluators, it produces results with the best accuracy and that are extremely similar to the scores provided by human educators.