Cheating detection analyzes patterns across student submissions to identify potential academic misconduct. While particularly valuable for online exams where traditional supervision is limited, the feature requires careful implementation to balance detection effectiveness with privacy concerns and false positive risks. Detection methods include:
- Answer Pattern Analysis: Identifying statistically improbable similarities in response sequences
- Time-based Analysis: Detecting suspicious patterns in submission timing and student behavior
- Collaboration Detection: Recognizing signs of unauthorized group work or shared answers
- Source Attribution: Identifying content potentially copied from unauthorized sources
The feature serves primarily as a screening tool to flag suspicious cases for human review, rather than making automated determinations. This aligns with our assisted grading approach and ensures compliance with requirements for high-risk systems.
Prerequisites
None
In use
Similar techniques are used in many online learning platforms.