Autenticação de alunos utilizando biometria comportamental em ambientes Juiz On-line

Abstract

In online judge environments, student authentication is usually done only at the beginning of the login session by typing a password, which, despite being the standard method, presents vulnerabilities, since, subsequently, the authenticity of the logged in user. In this way, sharing login and password information causes authenticity issues or identity spoofing. In the case of exercises and assessments carried out in environments like this, there is a need for non-intrusive verification of the students’ identity during the submission of the activities, not just at login. In this research, a method of authentication of students in online judge environments is being proposed and validated using behavioral biometrics, but specifically typing dynamics, this technique does not require any explicit action from the user, in addition to the low cost, as it is only necessary a conventional keyboard for acquiring information, making the cost relatively low. For this, a Siamese 1D CNN architecture was designed that learned, from the raw data of the users’ typing rhythm, the representation necessary for the recognition of the students, thus dispensing with the manual extraction of features, in which it is a process that requires a lot of time and the participation of experts. To validate this method, data from the typing dynamics of 42 students in the CodeBench online judge environment were used. Then, several experiments were carried out demonstrating the effectiveness of the proposed method. With the use of this method, as an additional module of security in programming environments, it has been proven through experiments that it is possible to reduce the probable occurrences of fraud on the part of students in programming exercises and evaluations. In this way, the application of the proposed method will provide institutions and teachers with additional security against fraud. In addition, it is important to the student as it testifies to their academic integrity in all activities. According to our literature review, the present work is the first method of authenticating students in online judge environments.