Advanced code similarity and AI-generated content detection
The Lumetrix Plagiarism Detection System uses multi-layered analysis to identify code similarities and detect AI-generated content. Our advanced algorithms provide comprehensive insights into code authenticity, helping educators maintain academic integrity.
Multi-layer structural comparison
Pattern recognition for generated code
Instant similarity scoring
Breaks down code into fundamental tokens and compares sequences. Effective at detecting direct copying with minor variable name changes.
Accuracy: High for direct plagiarism | Speed: Fast
Analyzes the structural representation of code, identifying similar logic patterns regardless of syntax variations. Catches restructured code plagiarism.
Accuracy: Very high for structural similarity | Speed: Moderate
Uses advanced algorithms to understand code meaning and purpose. Detects functionally equivalent code written differently, including algorithm reimplementations.
Accuracy: High for conceptual similarity | Speed: Moderate
Maps execution paths and decision structures within code. Identifies similar algorithmic approaches even with different implementations.
Accuracy: High for logic similarity | Speed: Fast
Detects characteristics typical of AI-generated code including comment patterns, variable naming conventions, error handling styles, and code organization.
Accuracy: High for AI detection | Speed: Moderate
Similarity score ≥ 80% or very high AI probability (≥ 85%)
Indicates strong evidence of plagiarism or AI-generated content. Requires immediate review and typically warrants investigation or academic action.
Similarity score 50-79% or moderate AI probability (60-84%)
Suggests possible plagiarism or AI assistance. Warrants closer examination and discussion with the student to understand the submission context.
Similarity score < 50% and low AI probability (< 60%)
Indicates original work with minimal similarity to other submissions. No immediate concerns, though some common patterns are expected in standard assignments.
Automatically compare all submissions within an assignment to detect similar patterns.
Paste and compare two code snippets directly for quick similarity analysis.
View layer-by-layer breakdown with confidence scores and specific matching segments.
Mark cases as reviewed, add notes, and track investigation progress.
Works with C, C++, Python, Java, JavaScript, and more programming languages.
All analysis happens securely, and student code is never shared or stored externally.
Regular Scanning: Run plagiarism checks on all assignments to establish baseline similarity patterns and catch issues early.
Context Matters: High similarity doesn't always mean plagiarism. Consider assignment requirements, teaching materials, and common patterns.
Use as a Tool: The detector is a screening tool, not a final judgment. Always review flagged cases manually before taking action.
Document Findings: Use the review notes feature to document your investigation and decisions for each flagged case.
Educational Approach: When discussing results with students, focus on learning outcomes and academic integrity rather than punishment.
The plagiarism detection system is exclusively available to teachers. Students do not have access to this feature to maintain the integrity of the detection process.