Development of a new approach to detect unwanted behavior in online exams

Distance learning is widely used by many educational institutions. Due to the pandemic caused by the coronavirus, the importance of distance learning has increased in our country as well as all over the world. With the transfer of educational activities to the online environment, unethical behaviors have also emerged. Behaviors such as cheating and plagiarism, especially in online exams, have become a source of concern for both learners and educational institutions. This thesis aims to reveal the best improved and optimized artificial intelligence models by predicting and classifying undesirable behavior patterns using a dataset consisting of artificial and real exam data of students taking online distance education courses in an online environment through a distance education system. Within the scope of the study, online exam data was analyzed by researchers using machine learning and deep learning algorithms to reveal the best model prediction and classification performance in regression and classification using two different scenarios. The model we proposed in Scenario-1 to detect students who "cheat" from undesirable behavior patterns is a four-layer DNN model. Within the scope of Scenario-2, the SVM model was found to be the best model we proposed in detecting students who "cheat" in the undesirable behavior analysis in binary classification and the RF model in ternary classification. In addition, explanatory methods were used to identify the factors that increase the performance of the best performing algorithms in the model performance. As a result of this study, it has been shown that the most appropriate parameter selection and application of layer functions that will improve performance can be effective in predicting complex problems that cannot be solved using target variables and classical mathematical models. The results prove that students' online distance education exam data can be easily applied to machine learning models and DNN models. The proposed models can provide educational institutions with a roadmap and insight in evaluating online exam applications and ensuring academic integrity.