A Novel Intelligent Assessment Technique in E-Learning for Subjective-Type Questions

Assessment is an essential component of every system of teaching and learning. Eventually, in a traditional evaluation method, the assessment is carried out by Subject Matter Experts (SMEs) who are human instructors. Regarding Industry 4.0 in addition to playing a big part, authentic learning is pursued in academics. For an automated evaluation to be conducted thoroughly, the system needs categorized objects at different cognitive domain levels. In their proposed method, pre-classified items which are of subjective type are used by the authors from the Item Bank. Each item is assigned a weighted value that relates to a certain cognitive domain level. The weights assigned to each item have been predetermined by the teachers or subject matter experts based on the students’ level of competency in that subject area and the pertinent level of the cognitive domain. The authors of this work created a system that uses various Machine Learning (ML) and Deep Learning (DL) approaches to deliver a subjective evaluation based on elements that have already been pre-classified. Random selection is used to select items which are already classified from the Item Bank that are distributed uniformly across the multiple levels of the perceptive/cognitive domain according to Bloom’s defined taxonomy. A pre-processing module is used to send each response that is received. To obtain the attributes, tokenization technique is used to the pre-processed answer that has been acquired. The retrieved features are fed into the regression module, which outputs them in a numerical format. The regression module assigns a classified answer to each evaluation item. A validation/training set that is readily available for the evaluation procedure is being compared with the indicated answers. A learner’s performance is evaluated using the above-described manner using metrics like accuracy.