Crowd-Sourced Supervisors for the Automatic Invigilation of Online Assessments

An increase in digitalisation and the compounded effects of the COVID-19 pandemic have forced educational institutions to adopt digital solutions for supervising online assessments. Misconduct in online assessments is increasing as institutions compromise academic integrity to remain operational. Implementing the right tools to mitigate the risk of academic dishonesty has become the priority in ensuring academic integrity. Existing proctoring tools are intrusive, less privacy-conscious and operate in a space that has limited to no standards. Due to the state of current proctoring tools, there is a lack of adequate supervision solutions, novel enough to deal with the issues of academic misconduct. This paper proposes an algorithm called Crowd-Vision, encapsulated in a web-based tool and powered by crowd-sourced supervisors, to decrease levels of academic dishonesty in online assessments. Crowd-Vision uses various configurable assessment parameters to simulate an assessment environment balanced with both real and generated invigilators. The evaluation of the web-based tool revealed that the tool has the potential to mitigate academic dishonesty.