Evaluation of cosine similarity and deci similarity on students’ test essay answers dataset

Learning evaluation is needed to determine the success of implementing educational institutions. Assessment of the learning process can be in the form of a short answer test,an essay, or multiple choice. As the main parameter, the exam must have an assessment rubic and competitive questions and become a significant part of the learning benchmark. The test results must be valid and reliable to determine whether the learning system has been appropriately implemented. In the digital era, it is necessary to have a fast and efficient test and have accuracy in giving short answer scores or essays. Teacher educators and lecturers will have a value gap when correction is made. Gaps are detrimental to students when answering questions from exams that have been carried out. Using cosine similarity and dice similarity is one solution to evaluating student learning. Cosine and dice similarity will compare the students’ answers with the answer key. The smaller the error value of the responses obtained, the participants’ answers get a high accuracy value. The value of the small error given by the two algorithms, cosine similarity in the short answers to the student exams, receives a score of 21%. 16.21% error value generated by dice similarity in quick answers.