The article considers options for implementing proctoring tools in distance learning systems based on information and communication technologies. A comparative analysis of such systems from leading developers in this field was conducted. A classification of approaches was made and a list of functional components providing a solution to the given task was formulated. It was noted that each of the solutions has its positive sides and disadvantages, based on which the requirements for the software for the ATutor e-learning system were formulated.
Distance learning forms of education have been around for quite some time, but the real surge in their popularity was due to the quarantine restrictions during the COVID-19 pandemic. This rapid development of distance learning technologies happened thanks to the use of modern information and communication technologies, which became its basis. And while the problem with individual assignments was solved more or less quickly, the remote assessment of knowledge by means of testing still raises many questions, starting with the security of personal data and ending with a significant number of opportunities to violate the knowledge assessment procedure.
An important task of developers and users of e-learning systems is the integration of automated means of proctoring into these systems, which would provide effective control over the knowledge assessment procedure. One of the technical components of the proctoring system, designed to solve this problem, are specialized means of identity verification during knowledge testing.
Modern proctoring systems are based on multimedia technical means widely used in distance education. The principle of their operation is commonly as follows: the user connects to the system through a browser, and the system confirms the identity in an automated mode, collects data from the microphone, web camera and, in some cases, from the computer screen during the knowledge assessment. At the same time, the system can automatically decide on access denial to the test or exam if the person recognized through the web camera does not match to the expected person or if they have violated other conditions of passing the test. Also, the collected data is provided for the review by the examiner, who makes final decision.
Common e-learning systems were initially developed without proctoring tools, and therefore the development of additional modules with proctoring features that would be able to integrate with these learning management systems (Learning management system, LMS) or distance learning systems is relevant. In some cases, integration is possible using a module already created by the system developers, and in others - using API that allows implementing integration on your own.
The goal of this study was to analyze the most widely used technical solutions for identity
verification in e-learning systems, to classify the main approaches and to choose the optimal option
for use in an e-learning system based on LMS ATutor for an educational institution [
Next, the most successful solutions to the problem of identity verification during knowledge assessment in distance learning conditions are considered, and their advantages and disadvantages are analyzed.
ExamOnline [
SpeedExam [
OES [
The system is able to integrate with such LMSs as Univer, Moodle, Canvas, Indigo, Sirius.
Moodle Proctoring [
Before starting the test, the module asks for permission to access the camera and permission to view the screen. After granting the permission, the user can see image of himself on the screen and start answering questions. At certain intervals, snapshots from the web camera and the screen are sent to the server, so the user can not try to do anything suspicious during assessment. Among the advantages: the solution is completely open (open source), integrates into LMS Moodle, thanks to which it works with the existing bank of test questions and the configured test solution of LMS Moodle; API for external access. By connecting the Amazon Rekognition or Brainstation Facerecognition API, it is possible to enable automatic face recognition. Among the disadvantages: it does not automatically make a decision on granting access to the test on its own - a human is needed to make a decision; there is no native implementation of face recognition algorithms - this function relies on third-party commercial APIs.
Proctortrack [
Proctorio [
Based on the results of the analysis of known existing solutions for identity verification during knowledge assessment, the following conclusions can be drawn: solutions like OES, Moodle Proctoring, Proctotrack, Proctorio allow to fully integrate them into common electronic learning systems (LMS) to use existing database of test questions and already configured tools for knowledge assessment, the rest do not; all solutions except Moodle Proctoring are commercial and closed source, and therefore do not allow any modification, customization and improvement; the considered technical solutions do not have the ability of direct integration with LMS
ATutor in any way; none of the solutions provide a guaranteed high percentage (95% or more) error-free result.
In identity verification, the main task is high-accuracy face recognition, which starts with face detection in an image, which is one of the variations of the general task of object detection. Face recognition can be defined as determining whether a given image contains faces, and if it does, finding the location of each face. Today, there are dozens of computer-based face detection and recognition methods. However, these methods do not provide 100% accuracy and reliability, also at the same time, often have limitations in recognition performance.
Among the main challenges and problems that arise in the implementation of human face recognition algorithms are: insufficient illumination conditions; variable expressions of emotions in the faces; different types of skin tones; varying distances from the camera to the face; varying face orientations; complex background, partial overlapping of faces with glasses, elements of clothing, hands, hair, medical masks, etc., insufficient resolution of video equipment, etc.
All these reasons determine the need for further research, with the aim of developing and implementing a technical solution using AI technologies, which would ensure automatic and effective verification of identity during knowledge assessment in the LMS of an educational institution, which would be based on the use of web technologies and common (including mobile) software and hardware of users that they use on daily basis. The desired solution should have at least 95% accuracy and perform 1 second per identification or better. It also should be able to integrate with LMS ATutor to use it’s test question database and knowledge assessment tools.