EUVIMEDEUVIMED entdecken
Open Access · CC BY 4.0

An Explainable AI-Driven Framework for Cyber-Legal Analytics, Digital Evidence Assessment, Cyber-Event Prediction, and Intelligent Regulatory Decision Support

Santosh Kumar Tripathi, Dr. Swati Mishra, Dr. Laishram Malem Mangal, Dr. Haider Ali

Computer and Decision Making: An International Journal · 05.11.2026 · 3 (6S) · 1899-1912

DOI: 10.1922/comdem.v3i6s.18957 ↗

Kurz gefasst

Fachartikel in der Zeitschrift Computer and Decision Making: An International Journal, online erschienen am 05.11.2026. Verfasst von Santosh Kumar Tripathi und 3 weiteren Autor:innen. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0).

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Cyber incidents are now resulting in data-disclosure, compliance and regulatory repercussions, but business decisions are not supported by transparency in prioritizing incidents for review. This study aims to create, implement and test an explainable artificial intelligence model to aid, rather than supplant, human regulatory and organizational decision making. A total of 10,596 incident records were validated from the public VERIS Community Database (VCDB), 13 exact duplicates were removed and 10,583 incidents were used for analysis. Three predictive tasks were analyzed—confirmed data disclosure as an incident assessment for use as evidence (usable N = 9,675); cyber-event/action-type classification (N = 8,540); and documented legal/regulatory impact (256 positive cases). Leakage-aware predictors were used, stratified 80/20 split was used and repeated stratified cross validation was used to compare the Logistic Regression, Random Forest and XGBoost. The two binary XGBoost models were explained by Native TreeSHAP. For confirmed disclosure, Random Forest and XGBoost exhibited similar good discrimination. For the cyber-event classification task, XGBoost was able to attain accuracy of 0.915 and macro-F1 of 0.871, albeit with less success in malware recall. The rare regulatory outcome achieved by XGBoost was 0.856 ROC-AUC and 0.258 PR-AUC, while the baseline prevalence was close to 0.024. The priority-review band, when ranked into percentile bands of held-out incidents, indicated about 17.0% observed regulatory impact as compared to an overall trend of about 2.4%. The results reveal that explainable ensemble models can be used to aid a risk-based triage for human review. They are not intended to express causation or legal liability, or to describe what is or is not admissible in court, but rather to describe predictive relationships between reported incidents

Zitieren

APA

Tripathi, S. K., Mishra, D. S., Mangal, D. L. M., & Ali, D. H. (2026). An Explainable AI-Driven Framework for Cyber-Legal Analytics, Digital Evidence Assessment, Cyber-Event Prediction, and Intelligent Regulatory Decision Support. Computer and Decision Making: An International Journal, 3(6S), 1899-1912. https://doi.org/10.1922/comdem.v3i6s.18957

Vancouver

Tripathi SK, Mishra DS, Mangal DLM, Ali DH. An Explainable AI-Driven Framework for Cyber-Legal Analytics, Digital Evidence Assessment, Cyber-Event Prediction, and Intelligent Regulatory Decision Support. Computer and Decision Making: An International Journal. 2026;3(6S):1899-1912. doi:10.1922/comdem.v3i6s.18957

BibTeX
@article{Tripathi2026,
  author = {Tripathi, Santosh Kumar and Mishra, Dr. Swati and Mangal, Dr. Laishram Malem and Ali, Dr. Haider},
  title = {An Explainable AI-Driven Framework for Cyber-Legal Analytics, Digital Evidence Assessment, Cyber-Event Prediction, and Intelligent Regulatory Decision Support},
  journal = {Computer and Decision Making: An International Journal},
  year = {2026},
  volume = {3},
  number = {6S},
  pages = {1899--1912},
  doi = {10.1922/comdem.v3i6s.18957}
}
Kostenlos starten →