MODEL DEVELOPMENT OF AUTOMATIC VALIDATION DETECTION ON SURVEY RESPONSES OF COURSE LEARNING PROCESSES USING BERT-BASED MODEL

Abstract

Evaluation of the course learning process is one of the important actions that must be carried out by educational institutions. Evaluation must be carried out thoroughly and in depth on what is felt by students. One form of evaluation that is usually given is in the form of a survey consisting of closed and open-ended questions. The process required to check the validity of student responses is quite time consuming if done manually. Therefore, in this study a model will be proposed to automatically check validation of survey responses to the course learning process using a BERT-based model. The dataset used is survey data taken from one of the educational institutions in Indonesia by considering the relation between closed and open-ended questions towards the same goal. The results obtained from this study are that all BERT-based models can provide a good accuracy value of above 80% in checking the validity of the responses given by students, when compared to manual judgement.