Resilient Anomaly Detection in Fiber-Optic Networks: A Machine Learning Framework for Multi-Threat Identification Using State-of-Polarization Monitoring
Open accessThe DOI matches its Crossref record; 47 of the 47 references it cites were found in a registry or an index; 5 of the 15 sources its claims cite were checked for support.
Abstract
We present a machine-learning framework based on real-time state-of-polarization (SOP) monitoring for anomaly identification in optical fiber networks, leveraging continuous SOP monitoring and angular speed (SOPAS) analysis across multiple physical vibration and perturbation scenarios.
Overview
CoChat Summary
TL;DR
This paper presents a supervised machine-learning framework using real-time state-of-polarization (SOP) monitoring to detect and classify fiber-optic threats, including vibrations, overlapping disturbances, and fiber tapping. Random forest and XGBoost achieve near-perfect clean-data accuracy and retain useful detection under superimposed noise, though performance degrades as noise frequency increases.
Key findings
- On clean data, random forest achieved 99.98% accuracy and F-1 0.9998; XGBoost 99.93% and 0.9993; decision tree 99.72% and 0.9972; k-NN 95.08% and 0.9507. Tree-based models reached AUC 1.00 on the clean dataset.
- For overlapping events, XGBoost was selected and predicted with 98% accuracy, misclassifying only 2.06% of overlap events.
- Under superimposed noise, random forest accuracy fell to 87.41% at 1 Hz, 70.50% at 3 Hz, and 58.99% at 5 Hz; corresponding F-1 scores were 0.873, 0.699, and 0.578. AUC for RF was 1.00 clean, 0.99 at 1 Hz, 0.90 at 3 Hz, and 0.92 at 5 Hz.
- Eavesdropping tests using an optical fiber identification device on a 13 km metropolitan cable showed bending-induced SOP changes, with the highest SOPAS angular speed reported as 2.5 rad/s; tests were repeated approximately 50 times.
Methods
- Experimental setup: a 1530 nm continuous-wave laser at 6 dBm launched into sensing fiber; two SMF segments of 8 km and 5 km plus an SMF section manipulated by an Arduino-controlled robotic arm; a Novoptel PM1000 polarimeter captured Stokes parameter evolution. Sampling was 1500 samples per second with averaging time exponent 16.
- Features: Stokes parameters S1, S2, S3 and SOPAS, plus rolling mean and standard deviation over windows of 500 and 1000, and lag features up to third order. Labels were No Event, Shaking-1 Hz, Shaking-3 Hz, Shaking-5 Hz, and Shaking-10 Hz; data were split 80:20 into training and testing.
- Classifiers and evaluation: k-NN (k in [5, 50]), decision tree (depths [3, 5, 10, 15, 20]), random forest (100 trees, max depth 6), and XGBoost (150 trees, learning rate 0.4). Evaluation used accuracy, precision, recall, F-1, confusion matrices, and one-vs-rest AUC-ROC; a weighted performance metric combined accuracy and normalized training time across weight configurations.
Limitations
- The paper states that future work is needed to improve resilience to environmental interference through adaptive filtering and noise-aware feature extraction, and that additional optimization is required for embedded and latency-sensitive applications via lightweight model tuning and latency-aware feature refinement.
- The paper states that noise-mitigation techniques are necessary for real-world implementations.
- Evident from the text: robustness testing used superimposed sinusoidal noise rather than measured field interference, and the vibration experiments were conducted in a controlled testbed, which may limit direct generalization to real deployments.
Model-written summary. Check it against the paper before relying on it.
Integrity signals
Registry
Cites
Statistics
Verification
Claims the paper makes
4 Valid · 4 Not valid · 7 Not checked · 86 Own results and interpretations
The sources with a verdict support this claim; the rest weren’t checked yet, have no full text we could read, or their verdicts aren’t rated yet.
In the paper
“Utilizing XGBoost [41], our model accurately classified overlapping fiber anomalies, which are further discussed in [30].”
What the sources say
- Not checked — no full text of the sourcecites [41] 41. Zhang, C.; Wang, D.; Wang, L.; Guan, L.; Yang, H.; Zhang, Z.; Chen, X.; Zhang, M. Cause-aware failure detection using an interpretable XGBoost for optical networks. Opt. Express 2021 , 29 , 31974-31992. [CrossRef] [PubMed] source
- Supportedcites [30] 30. Malik, G.; Masood, M.U.; Dipto, I.C.; Mohamed, M.C.; Straullu, S.; Bhyri, S.K.; Galembirti, G.M.; Pedro, J.; Napoli, A.; Wakim, W.; et al. Intelligent Detection of Overlapping Fiber Anomalies in Optical Networks Using Machine Learning. In Proceedings of the IEEE Summer Topicals, Berlin, Germany, 21-23 July 2025. source
The Introduction states that the proposed XGBoost model achieves high accuracy in classifying overlapping fiber anomalies. Section III reports that only 2.06% of overlapping events were misclassified in testing, supporting the attributed performance claim. (LLM, conf=0.94)
plainstates a cause- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
A result the paper reports from its own data — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“No noise | 99.98 | 99.93 | 95.08 | 99.72 | 0.9998 | 0.9993 | 0.9507 | 0.9972”
plain- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: unclear — It is unclear whether the paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
A result the paper reports from its own data — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“Random forest emerged as the most accurate classifier, achieving a near-perfect accuracy of 99.98%, followed by XGBoost and decision tree, while k-NN lagged with an accuracy of 95.08%.”
plain- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: unclear — It is unclear whether the paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
A result the paper reports from its own data — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“After saving and testing the ideal model on the datasets with noise, it was noticed that further enhancements to these features were necessary for that model to perform better on the datasets with varying noise levels.”
plainstates a cause- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
A result the paper reports from its own data — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“The instantaneous power gradually changes between levels over a period of about 1 s when the clamp is closed, exhibiting a sudden spike, respectively.”
plainstates a cause- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: unclear — It is unclear whether the paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
61 more claims
Create a free account to see all 101 claims and every source verdict.
- It may state two separate claims as one. 19
- Probably not a substantial claim. 18
- The quote only partly states this claim. 1
Self-check
Percentages, sums and reported differences re-computed from the paper’s own numbers.
Each contradiction between two passages listed here was confirmed by an independent check.
Rated minor: it would not touch a result — a value no result rests on, a p-value that stays on the same side of the significance level, or a conflict that is clerical, resolved elsewhere in the paper, or outside its own findings.
In the paper
§ 2.1. Eavesdropping
“After 30 s of gentle clamping, the fiber was released.”
§ 2.1. Eavesdropping
“The outcomes of six tests are illustrated in Figure 1, where the behavior of the stokes parameters is evident for a clamp held for approximately 10 s.”
Rated minor: it would not touch a result — a value no result rests on, a p-value that stays on the same side of the significance level, or a conflict that is clerical, resolved elsewhere in the paper, or outside its own findings.
In the paper
§ 5.1. Performance Evaluation of Model Classification Scores
“The resulting classification accuracies were 87.41%, 70.50%, and 58.99%, respectively.”
§ 5.2. Performance Evaluation of ML Models Using Weighted Metrics
“| 1 Hz noise | 87.5 | - | - | - | 0.873 | - | - | - |”
Rated minor: it would not touch a result — a value no result rests on, a p-value that stays on the same side of the significance level, or a conflict that is clerical, resolved elsewhere in the paper, or outside its own findings.
In the paper
§ 5.1. Performance Evaluation of Model Classification Scores
“However, RF was better at distinguishing between the four labels in this experiment.”
§ 5.1. Performance Evaluation of Model Classification Scores
“These matrices illustrate how well each classifier predicted the five distinct event categories: No Event, and Shaking at 1 Hz, 3 Hz, 5 Hz, and 10 Hz.”
Citations
Cites
Gradient-based learning applied to document recognition
Lecun, Y., Bottou, L., Bengio, Y., & Haffner, P. (1998) · Proceedings of the IEEE · doi:10.1109/5.726791 · Cited by 58,805
Zhang, W., Li, C., Peng, G., Chen, Y., & Zhang, Z. (2018) · Mechanical Systems and Signal Processing · doi:10.1016/j.ymssp.2017.06.022 · Cited by 1,421
Introduction to machine learning: k-nearest neighbors
Zhang, Z. (2016) · Annals of Translational Medicine · doi:10.21037/atm.2016.03.37 · Cited by 1,294
Collett, E. (2005) · SPIE eBooks · doi:10.1117/3.626141 · Cited by 551
Learning decision tree classifiers
Quinlan, J. R. (1996) · ACM Computing Surveys · doi:10.1145/234313.234346 · Cited by 504
The Critical Node Detection Problem in networks: A survey
Lalou, M., Tahraoui, M. A., & Kheddouci, H. (2018) · Computer Science Review · doi:10.1016/j.cosrev.2018.02.002 · Cited by 311
Optical Fiber Sensors and Sensing Networks: Overview of the Main Principles and Applications
Pendão, C., & Silva, I. (2022) · Sensors · doi:10.3390/s22197554 · Cited by 305
City-Scale Dark Fiber DAS Measurements of Infrastructure Use During the COVID-19 Pandemic
Lindsey, N. J., Yuan, S., Lellouch, A., Gualtieri, L., Lecocq, T., & Biondi, B. (2020) · Geophysical Research Letters · doi:10.1029/2020gl089931 · Cited by 162
Machine-learning-based anomaly detection in optical fiber monitoring
Abdelli, K., Cho, J. Y., Azendorf, F., Griesser, H., Tropschug, C., & Pachnicke, S. (2022) · Journal of Optical Communications and Networking · doi:10.1364/jocn.451289 · Cited by 149
Polarization sensing using submarine optical cables
Mecozzi, A., Cantono, M., Castellanos, J. C., Kamalov, V., Muller, R., & Zhan, Z. (2021) · Optica · doi:10.1364/optica.424307 · Cited by 131
Seismic Arrival-time Picking on Distributed Acoustic Sensing Data using Semi-supervised Learning
Zhu, W., Biondi, E., Li, J., Yin, J., Ross, Z. E., & Zhan, Z. (2023) · Nature Communications · doi:10.1038/s41467-023-43355-3 · Cited by 130
Chen, X., Li, B., Proietti, R., Zhu, Z., & Yoo, S. J. B. (2019) · Journal of Lightwave Technology · doi:10.1109/jlt.2019.2902487 · Cited by 107
Optical fiber tapping: Methods and precautions
Zafar Iqbal, M., Fathallah, H., & Belhadj, N. (2011) · 8th International Conference on High-capacity Optical Networks and Emerging Technologies · doi:10.1109/honet.2011.6149809 · Cited by 77
Proactive Fiber Damage Detection in Real-time Coherent Receiver
Boitier, F., Lemaire, V., Pesic, J., Chavarria, L., Layec, P., Bigo, S., & Dutisseuil, E. (2017) · 2017 European Conference on Optical Communication (ECOC) · doi:10.1109/ecoc.2017.8346077 · Cited by 68
Natalino, C., Schiano, M., Di Giglio, A., Wosinska, L., & Furdek, M. (2019) · Journal of Lightwave Technology · doi:10.1109/jlt.2019.2923558 · Cited by 63
Cause-aware failure detection using an interpretable XGBoost for optical networks
Zhang, C., Wang, D., Wang, L., Guan, L., Yang, H., Zhang, Z., Chen, X., & Zhang, M. (2021) · Optics Express · doi:10.1364/oe.436293 · Cited by 53
Reflective fiber fault detection and characterization using long short-term memory
Abdelli, K., Grießer, H., Ehrle, P., Tropschug, C., & Pachnicke, S. (2021) · Journal of Optical Communications and Networking · doi:10.1364/jocn.423625 · Cited by 53
Failure Management Overview in Optical Networks
Cruzes, S. (2024) · IEEE Access · doi:10.1109/access.2024.3498704 · Cited by 49
Support Vector Machines for Anomaly Detection
Zhang, X., Gu, C., & Lin, J. (2006) · 2006 6th World Congress on Intelligent Control and Automation · doi:10.1109/wcica.2006.1712831 · Cited by 48
Predicting the actual location of faults in underground optical networks using linear regression
Nyarko‐Boateng, O., Adekoya, A. F., & Weyori, B. A. (2021) · Engineering Reports · doi:10.1002/eng2.12304 · Cited by 35
Distributed fiber-optic sensor for detection and localization of acoustic vibrations
Sifta, R., Munster, P., Sysel, P., Horvath, T., Novotny, V., Krajsa, O., & Filka, M. (2015) · Metrology and Measurement Systems · doi:10.1515/mms-2015-0009 · Cited by 33
Liu (2023) · doi:10.1145/3596262 · Cited by 31
Theory of phase transmission fibre-optic deformation sensing
Fichtner, A., Bogris, A., Nikas, T., Bowden, D., Lentas, K., Melis, N. S., Simos, C., Simos, I., & Smolinski, K. (2022) · Geophysical Journal International · doi:10.1093/gji/ggac237 · Cited by 29
Algorithm optimization for rockfalls alarm system based on fiber polarization sensing
Pellegrini, S., Rizzelli, G., Barla, M., & Gaudino, R. (2023) · IEEE Photonics Journal · doi:10.1109/jphot.2023.3281670 · Cited by 21
Machine-learning-based method for fiber-bending eavesdropping detection
Song, H., Lin, R., Li, Y., Lei, Q., Zhao, Y., Wosinska, L., Monti, P., & Zhang, J. (2023) · Optics Letters · doi:10.1364/ol.487214 · Cited by 21
Anomaly Detection and Localization in Optical Networks Using Vision Transformer and SOP Monitoring
Abdelli, K., Lonardi, M., Gripp, J., Correa, D., Olsson, S., Boitier, F., & Layec, P. (2024) · Optical Fiber Communication Conference (OFC) 2024 · doi:10.1364/ofc.2024.tu2j.4 · Cited by 19
Sadighi, L., Karlsson, S., Natalino, C., & Furdek, M. (2024) · Optical Fiber Communication Conference (OFC) 2024 · doi:10.1364/ofc.2024.m1h.1 · Cited by 19
Deep Learning for Detection of Harmful Events in Real-World, Noisy Optical Fiber Deployments
Sadighi, L., Karlsson, S., Natalino, C., Wosinska, L., Ruffini, M., & Furdek, M. (2025) · Journal of Lightwave Technology · doi:10.1109/jlt.2025.3557748 · Cited by 18
Enhancing fiber security using a simple state of polarization analyzer and machine learning
Tomasov, A., Dejdar, P., Munster, P., Horvath, T., Barcik, P., & Da Ros, F. (2023) · Optics & Laser Technology · doi:10.1016/j.optlastec.2023.109668 · Cited by 17
Bao, Y., Chen, G., Meng, W., et al. (2015) · doi:10.1115/jrc2015-5653 · Cited by 15
Abdelli, K., Lonardi, M., Gripp, J., Olsson, S., Boitier, F., & Layec, P. (2023) · IET Conference Proceedings · doi:10.1049/icp.2023.2374 · Cited by 13
Sadighi, L., Karlsson, S., Wosinska, L., & Furdek, M. (2024) · 2024 24th International Conference on Transparent Optical Networks (ICTON) · doi:10.1109/icton62926.2024.10648140 · Cited by 13
Yilmaz, A. K., Deniz, A., & Yuksel, H. (2021) · 2021 International Conference on Electrical, Computer and Energy Technologies (ICECET) · doi:10.1109/icecet52533.2021.9698452 · Cited by 11
Spurny, V., Dejdar, P., Tomašov, A., Münster, P., & Horváth, T. (2023) · International Workshop on Fiber Optics in Access Networks (FOAN) · doi:10.1109/foan59927.2023.10328134 · Cited by 11
Fiber-optic detection of snow avalanches using telecommunication infrastructurePreprint
Edme, P., Paitz, P., Walter, F., van Herwijnen, A., & Fichtner, A. (2023) · arXiv (Cornell University) · doi:10.48550/arxiv.2302.12649 · Cited by 11
A Machine Learning-Driven Smart Optical Network Grid for Earthquake Early Warning
Awad, H., Usmani, F., Virgillito, E., Bratovich, R., Proietti, R., Straullu, S., Pastorelli, R., & Curri, V. (2024) · International Conference on Transparent Optical Networks · doi:10.1109/icton62926.2024.10648206 · Cited by 10
Self-citationMachine Learning for Predictive Multi-Event Detection in Fiber Optic Systems
Malik, G., Masood, M. U., Mohamed, M. C., Straullu, S., Bhyri, S. K., Galimberti, G. M., Pedro, J., Napoli, A., Wakim, W., & Curri, V. (2025) · IEEE International Conference on Machine Learning for Communication and Networking (ICMLCN) · doi:10.1109/icmlcn64995.2025.11140455 · Cited by 9
Self-citationSadighi, L., Karlsson, S., Natalino, C., Wosinska, L., Ruffini, M., & Furdek, M. (2024) · Chalmers Research (Chalmers University of Technology) · Cited by 8
Predicting the Actual Location of Faults in Underground Optical Networks using Linear RegressionPreprint
Boateng, O. N., Adekoya, A. F., & Weyori, B. A. (2020) · doi:10.22541/au.158888207.73369927 · Cited by 4
Intelligent Detection of Overlapping Fiber Anomalies in Optical Networks Using Machine Learning
Malik, G., Dipto, I. C., Masood, M. U., Mohamed, M. C., Straullu, S., Bhyri, S. K., Galimberti, G. M., Pedro, J., Napoli, A., Wakim, W., et al. (2025) · 2025 IEEE Photonics Society Summer Topicals Meeting Series (SUM) · doi:10.1109/sum65312.2025.11121809 · Cited by 3
Self-citationLei, Q., Li, Y., Song, H., Wang, W., Zhao, Y., Zhang, J., & Liu, Y. (2023) · 2023 Opto-Electronics and Communications Conference (OECC) · doi:10.1109/oecc56963.2023.10209623 · Cited by 3
Principles of fiber optic sensors
Tosi, D., Sypabekova, M., Bekmurzayeva, A., Molardi, C., & Dukenbayev, K. (2022) · Optical Fiber Biosensors · doi:10.1016/b978-0-12-819467-6.00016-0 · Cited by 3
ML-based Anomaly Detection in Optical Fiber MonitoringPreprint
Abdelli, K., Cho, J. Y., & Tropschug, C. (2022) · arXiv (Cornell University) · doi:10.48550/arxiv.2202.11756 · Cited by 3
Malik, G., Ahmad, A., Ahmad, A., et al. (2021) · doi:10.1364/acpc.2021.t4a.273 · Cited by 1
Self-citationAbdelli, K., Lonardi, M., Gripp, J., Olsson, S., Boitier, F., & Layec, P. (2024) · 2024 24th International Conference on Transparent Optical Networks (ICTON) · doi:10.1109/icton62926.2024.10647504 · Cited by 1
Predicting the Actual Location of Faults in Underground Optical Networks using Linear Regression
Adekoya, A., Nyarko-Boateng, O., & Weyori, B. (2020) · doi:10.22541/au.159526966.66568186
Research works printed but not linked: 1
- 28. Malik, G.; Masood, M.U.; Dipto, I.C.; Mohamed, M.C.; Straullu, S.; Bhyri, S.K.; Galembirti, G.M.; Pedro, J.; Napoli, A.; Wakim, W.; et al. SOP-Based Anomaly Detection Leveraging Machine Learning for Proactive Optical Restoration. In Proceedings of the Optical Network Design and Modelling ONDM, Pisa, Italy, 6-9 May 2025.
Cited by
AI-Driven Fault Prediction and Restoration Leveraging Real-Time SOP Monitoring
Malik, G., Dipto, I. C., Masood, M. U., Ali, A., Mohamed, M. C., Straullu, S., Bhyri, S. K., Galimberti, G. M., Pedro, J., Napoli, A., et al. (2025) · International Conference on Software, Telecommunications and Computer Networks (SoftCOM) · doi:10.23919/softcom66362.2025.11197358 · Cited by 8
“We conducted a comprehensive evaluation of multiple supervised learning algorithms to identify the most suitable model for our anomaly classification task in [15].”
Self-citationMulti-tech sensing in-field demonstration on terrestrial optical data networks
Virgillito, E., Notarstefano, F., Ambrosone, R., et al. (2026) · Journal of Optical Communications and Networking · doi:10.1364/jocn.590101 · Cited by 1
“Finally, we report on controlled experiments using a robotic arm to simulate mechanical perturbations, validating the system’s ability to detect and characterize complex urban anomalies [30].”
Self-citationStrelnikov*, I. (2026) · Australian Journal of Engineering and Innovative Technology · doi:10.34104/ajeit.026.03460353
Malik, G., Masood, M. U., Ambrosone, R., Ali, A., Mohamed, M. C., Straullu, S., Bhyri, S. K., Galimberti, G. M., Pedro, J., Napoli, A., et al. (2026) · International Conference on Optical Network Design and Modeling (ONDM) · doi:10.23919/ondm68511.2026.11618868
Self-citationMachine Learning-Based Detection and Classification of Overlapping Fiber Anomalies
Malik, G., Masood, M. U., Ali, A., Mohamed, M. C., Straullu, S., Nespola, A., Bhyri, S. K., Napoli, A., Pedro, J., Galimberti, G. M., et al. (2026) · IEEE Photonics Technology Letters · doi:10.1109/lpt.2026.3678082 · Cited by 1
“As an early warning, the architecture should be able to classify the type of anomaly and track it to mitigate its impact, as much as possible, by implementing preventive steps [6].”
Self-citation
Metadata
- Authors
- Malik, G., Dipto, I. C., Masood, M. U., Mohamed, M. C., Straullu, S., Bhyri, S. K., Galimberti, G. M., Napoli, A., Pedro, J., Wakim, W., & Curri, V.
- Affiliations
- Politecnico di Torino (IT)
- LINKS Foundation (IT)
- Nokia (Finland) (FI)
- Corresponding author
- Malik, Gulmina
- ORCID
- Venue
- AI
- Publisher
- MDPI AG
- Published
- 2025-06-20
- Volume
- 6
- Issue
- 7
- Pages
- 131
- ISSN
- 2673-2688 (ISSN-L)
- OpenAlex
- W4411498779
- Cited by
- 10
- References
- 44
- Citations by year
- 2026: 8 · 2025: 2
- FWCI
- 2.22
- Type
- journal article
- Language
- en
- Open access
- gold
- doi.org
- Licence
- Locations
- AI (publishedVersion, cc-by)
- Preprints.org (acceptedVersion, cc-by)
- PORTO@Iris (submittedVersion, cc-by)
- DOAJ (DOAJ: Directory of Open Access Journals) (submittedVersion, cc-by-sa)
- Funders
- project PNRR-NGEU (MUR–DM117/2023, 101092766, 101119983)
- European Union’s Horizon Europe research and innovation program (MUR–DM117/2023, 101092766, 101119983)
- MSCA-DN Nestor GA (MUR–DM117/2023, 101092766, 101119983)
- Topics
- Physical Sciences › Engineering › Electrical and Electronic Engineering › Optical Network Technologies
- Physical Sciences › Engineering › Electrical and Electronic Engineering › Advanced Fiber Optic Sensors
- Physical Sciences › Computer Science › Computer Networks and Communications › Network Security and Intrusion Detection
- Keywords
- anomaly detection, vibration sensing, eavesdropping, mechanical perturbation, noise mitigation, real-time intrusion detection
About these checks
Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.
The claim verdicts this report counts are experimental: their accuracy is not yet measured.
A language model drew the claims from the paper on 2026-10-08, under version 3 of the claim definition (model deepseek-v4.1-flash, prompt extract-v3).
A second model labelled each claim on 2026-10-08 (model jev-1.13-20260917, definition revision gold-guideline-v0.2).
Cited sources are checked with a language model’s help (gpt-6-sol).
Statistics and arithmetic are recomputed, not judged.
Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.
What counts as a claim ›How we measure confidence ›How we check this ›
Are you an author, or do you see a problem with this page? Request a review
Data sources
Bibliographic data: Semantic Scholar, OpenAlex, Crossref.
Contains information from the Semantic Scholar Academic Graph Datasets, made available under the ODC Attribution License (ODC-BY 1.0).
Semantic Scholar’s data is described in: Kinney, R., et al. (2023). The Semantic Scholar Open Data Platform. arXiv:2301.10140.