System for online automated exam proctoring
Certain embodiments may be directed to a system and method for proctoring an exam, and more particularly, for online automated exam proctoring. A method may include checking a test-taker's computing device for compatibility, and for content that provide unauthorized aid to the test-taker during a testing session. The method may also include taking a photo of the test-taker, recording the test-taker by performing a room pan while analyzing the surrounds of the test-taker for unauthorized objects, and validating the identity of the test-taker by way of at least one of presenting challenge questions, obtaining voice biometrics, or obtaining keystroke biometrics. The method may further include recording the test-taker's audio or video and desktop feed, determining if the test-taker is exhibiting questionable behavior, and suspending the testing session if it is determined that the test-taker is exhibiting questionable behavior.
1 . A method, comprising:
controlling, via a processing server, a remote computer to administer an automated exam to a test-taker, the remote computer having a plurality of input devices;
receiving by the Processing server from the remote computer one or more streams of biometric information about the test-taker, a photo ID of the test-taker, a confirmation of exam rules by the test-taker, and answers to challenge questions by the test-taker collected from the input devices prior to the exam and/or during the exam;
establishing, by the processing server via implementation of supervised machine learning a threshold for normal by defining a baseline of an environment of the test-taker prior to the exam based on the one or more streams of biometric information, lighting of the environment, sound of the environment, movement of the test-taker, input from an institution, and input from an instructor, wherein establishing the threshold for normal comprises calibrating the processing server to operate within an acceptable predefined limit of the baseline;
recording, by the Processing server via the remote computer, biometric information from the one or more streams of biometric information, audio data, and video data of the test-taker during the exam;
comparing, by the processing server via implementation of supervised machine learning, the recorded biometric information, the recorded audio data, and the recorded video data to the baseline, and recognizing patterns of questionable behavior outside the threshold for normal;
determining, by the processing server by implementation of the supervised machine learning, that the test-taker is exhibiting questionable behavior based on the recognized patterns of questionable behavior outside the threshold for normal;
stopping, by the Processing server, the exam when it is determined that the test-taker is exhibiting questionable behavior;
in response to stopping the exam, flagging, by the processing server, the exam for further review by a live human proctor;
in response to the flagging, enabling the live human proctor to communicate remotely with the test-taker via a computer associated with the live human proctor and the remote computer by a chat feature or voice communication, review why the exam was stopped and either confirm or deny the determined questionable behavior, and then allow the test-taker to continue the exam or end the exam based on their review; and
automatically refining and updating, by the processing server, using the supervised machine learning implemented on the processing server, the initial pre-programmed behavioral patterns based on the accuracy of the recognized pattern of questionable behavior according to the subsequent confirmation or denial by the live human proctor.
2 . The method according to claim 1 , wherein the recorded biometric information and the audio or video data are reviewed by the human proctor after the exam has been completed to detect changes in the biometric information, the audio data, and the video data.
3 . The method of claim 1 ,
wherein the plurality of input devices includes a camera,
wherein the method further comprises issuing an instruction to cause the camera to obtain captured image data of the surroundings of the test-taker, and
wherein the method further comprises
detecting unauthorized objects in the surroundings of the test taker, and
suspending the examination if the unauthorized objects are detected in the captured image data.
4 . The method according to claim 1 , further comprising:
issuing an instruction to cause the camera to pan the surroundings of the test-taker.
5 . The method of claim 1 , further comprising:
instructing the remote computer to issue a warning that the test-taker has been detected of exhibiting questionable behavior.