As we go towards automation, a system for automatically evaluating descriptive responses is now required. Manual evaluation is time- and labor-intensive. Currently, our automated methods for descriptive, one-sentence, and objective responses are less accurate. The automatic scoring of answer scripts has shown to be useful in our experiments, and frequently the scores are assigned match the marks that are personally assessed. The development of an automated answer evaluation system based on machine learning is the aim of this study. The system will count the words and letters in the text that were retrieved from the pre-processed data in order to evaluate the response. The next step is to implement Natural Language Processing (NLP) to sanitize the retrieved text. Automated answer assessment is a crucial component. So, as we go towards automation, we need a framework for automatically evaluating descriptive answers. Manual evaluation requires a lot of time and effort. For objective-type, one-sentence, and descriptive answers, we currently have automated systems with lower accuracy. In our trials, the automatic scoring of answer scripts has shown to be beneficial, and frequently, the scores assigned coincide with the marks that are manually assessed. In this study, we aim to develop an automated system based on machine learning. The pre-processed data will be used to retrieve text from which the system will count the words and characters in order to evaluate the response. Natural Language Processing (NLP) must then be implemented in order to clean the retrieved text. An essential component is the automatic answer evaluation.