Presentation Title

Uniproctor: Ensuring Integrity of Online Exams Through Consolidation of Remote Live and Biometrics Based Proctoring Methods

Start Date

November 2016

End Date

November 2016

Location

Watkins 2141

Type of Presentation

Oral Talk

Abstract

With the popularity of online education, strategies are needed to ensure the academic integrity in remotely administered online exams. In traditional settings, students take exams in proctored classrooms. Remote online exams, however, are typically taken in uncontrolled locations (e.g., homes, offices, and libraries,) where cheating is easier.

We are developing Uniproctor, a system that streams the video of the test taker and their desktop to our biometrics software, which analyzes the video for signs of cheating. Example of signs include the student leaving the exam, being replaced by another individual, or having another person, a potential cheating accomplice, appear in the camera view.

The same video is streamed to the live proctor who sees the test taker, their desktop, and the cues from the biometrics software. Since a proctor typically watches multiple test takers, biometric cues inform the proctor of specific events triggered by a test taker, allowing the proctor to investigate the event for potential cheating. The proctor analyzes the legitimacy of the biometric cues because biometrics lacks the intelligence to contextualize events (e.g., flagging an innocent passerby in a public library, where the student takes an exam, as a potential cheating accomplice).

Overall, the system delivers a more effective and accurate approach to detecting cheating than the current approaches based solely on live proctors or biometrics. Uniproctor combines the pattern recognition power of biometrics and the discerning power human intelligence in order to best detect cheating in online exams.

This document is currently not available here.

Share

COinS
 
Nov 12th, 2:45 PM Nov 12th, 3:00 PM

Uniproctor: Ensuring Integrity of Online Exams Through Consolidation of Remote Live and Biometrics Based Proctoring Methods

Watkins 2141

With the popularity of online education, strategies are needed to ensure the academic integrity in remotely administered online exams. In traditional settings, students take exams in proctored classrooms. Remote online exams, however, are typically taken in uncontrolled locations (e.g., homes, offices, and libraries,) where cheating is easier.

We are developing Uniproctor, a system that streams the video of the test taker and their desktop to our biometrics software, which analyzes the video for signs of cheating. Example of signs include the student leaving the exam, being replaced by another individual, or having another person, a potential cheating accomplice, appear in the camera view.

The same video is streamed to the live proctor who sees the test taker, their desktop, and the cues from the biometrics software. Since a proctor typically watches multiple test takers, biometric cues inform the proctor of specific events triggered by a test taker, allowing the proctor to investigate the event for potential cheating. The proctor analyzes the legitimacy of the biometric cues because biometrics lacks the intelligence to contextualize events (e.g., flagging an innocent passerby in a public library, where the student takes an exam, as a potential cheating accomplice).

Overall, the system delivers a more effective and accurate approach to detecting cheating than the current approaches based solely on live proctors or biometrics. Uniproctor combines the pattern recognition power of biometrics and the discerning power human intelligence in order to best detect cheating in online exams.