Plagiarism Detector is not a black box. Its detection engine has been entered into the field's primary scientific benchmark — the PAN competition — and judged by independent academic organizers across nine international evaluations. The complete record is below; every result links to its official source.
| Year | Competition | Venue | Task | Rank | Score | Source |
|---|---|---|---|---|---|---|
| 2009 | 1st International Competition on Plagiarism Detection | PAN @ SEPLN · San Sebastián | External detection | 4 / 10 | 0.3045 | overview |
| 2010 | 2nd International Competition on Plagiarism Detection | PAN @ CLEF · Padua | External detection | 8 / 18 | 0.5093 | overview · paper |
| 2011 | 3rd International Competition on Plagiarism Detection | PAN @ CLEF · Amsterdam | External (WordNet experiment) | 7 / 9 | 0.19 | overview · paper |
| 2011 | CL!TR — Cross-Language Indian Text Re-Use | PAN@FIRE · IIT Bombay, Mumbai | Hindi↔English reuse | 4 (best run) | F 0.603 | leaderboard · overview |
| 2012 | 4th International Competition on Plagiarism Detection | PAN @ CLEF · Rome | Text alignment | 6 / 10 | 0.538 | overview · paper |
| 2012 | CL!NSS — Cross-Language Indian News Story Search 1ST | PAN@FIRE · ISI Kolkata | Journalistic reuse (en→hi) | 1 / 3 | NDCG@10 0.34 | overview · paper |
| 2013 | 5th International Competition on Plagiarism Detection | PAN @ CLEF · Valencia | Text alignment | 6 / 9 | 0.61523 | overview · paper |
| 2014 | 6th International Competition on Plagiarism Detection 1ST* | PAN @ CLEF · Sheffield | Text alignment | 1 std · 3/10 | 0.868 | overview · paper |
| 2026 | Voight-Kampff Generative AI Detection TOP AUC | PAN @ CLEF · Jena | Human-vs-AI text | 3 / 34 | ROC-AUC 0.996 | leaderboard |
plagdet = the PAN overall plagiarism-detection score (precision, recall and granularity combined). NDCG@10 = ranking quality. ROC-AUC = threshold-independent classification quality. *2014: 1st on the standard test corpus, 3rd of 10 on the official ranking corpus. 2026: 3rd of 34 teams; highest ROC-AUC of any team or baseline.
Each competition documented a technique that informs the shipping engine. The research and the software are one lineage — not marketing claims, but published, peer-reviewed methods.
Fast candidate detection of reused passages (PAN 2010, 2014).
Precise matching of source ↔ suspicious passages with high precision (PAN 2012–2014).
WordNet and TF-IDF / translation methods for paraphrase and cross-lingual reuse (PAN 2011; FIRE 2011–2012).
Distinguishing human from machine authorship — top ROC-AUC at PAN 2026.
Stand with Ukraine!