Machine Learning versus Simple Rule-Based Baselines for Phishing Website Detection: Effects of Domain-Grouped Evaluation and Feature Sources | IJCT Volume 13 – Issue 6 | IJCT-V13I6P97

IJCT
International Journal of Computer Techniques
ISSN 2394-2231 · Peer-Reviewed · Open Access
📚 Volume 13, Issue 6
📅 November 18, 2026
📄 Pages 819–823
🔖 ID: IJCT-V13I6P97

Machine Learning versus Simple Rule-Based Baselines for Phishing Website Detection: Effects of Domain-Grouped Evaluation and Feature Sources

Author(s)

Sharv Magdum, Aditya Jadhav, Aniket Khandagale, Harshal Pawar

Abstract

Phishing detectors are often judged by random-split accuracy on public benchmarks, against weak baselines or none. Using the Hannousse and Yahiouche dataset (11,430 URLs, 87 features), we examine how much machine learning (ML) improves on simple rule-based baselines, how much of that accuracy survives when related URLs are kept apart, and which feature groups carry it. We compare a six-rule URL heuristic, a domain blacklist, a one-split decision stump and shallow decision trees with Logistic Regression, Random Forest and XGBoost, using repeated 5-fold cross-validation (15 folds) with random folds and with class-balanced folds grouped by registered domain. XGBoost on all features reached 97.08 ± 0.32% accuracy with random folds and 95.17 ± 1.06% with domain-grouped folds, against 64.3% for the rule heuristic and 78.1% for the blacklist (random folds only; zero recall by construction in grouped folds). The strongest simple baseline, a one-split stump on the external feature google_index, reached 86.5%. Domain grouping lowered tree-model accuracy by 1.6 to 1.9 points with all features and by 3.9 to 4.2 points with URL-only features, and reduced XGBoost’s true-positive rate at 1% false positives from 92.7% to 83.1%. External-service features were the most informative group, and URL-only XGBoost (87.5% on grouped folds) was not significantly better than the stump. A dataset audit shows sampling artifacts. ML clearly helps, but part of the apparent gain reflects baseline choice, dataset sampling and external lookups

Keywords

phishing detection, machine learning, evaluation, domain-grouped cross-validation, feature ablation, dataset bias.

Conclusion

On this benchmark, XGBoost with all features outperformed every simple baseline, reaching 97.08% (random folds) and 95.17% (domain-grouped folds) accuracy. The advantage is smaller against a one-split stump (8.7 to 10.6 points) and disappears for URL-only models on grouped folds. Registered-domain grouping reduced accuracy by up to about 4 points and the true-positive rate at 1% FPR by up to about 18 points, and external features carried much of the accuracy while the audit points to dataset sampling artifacts. Future work should add time-split and second datasets, published and real-blocklist baselines, tuning and calibration, adversarial tests and measurements of lookup latency.

References

[1] S. Sheng, B. Wardman, G. Warner, L. Cranor, J. Hong, and C. Zhang, “An empirical analysis of phishing blacklists,” in Proc. Conf. Email and Anti-Spam (CEAS), 2009, doi: 10.1184/r1/6469805.
[2] A. M. Butnaru, A. Mylonas, and N. Pitropakis, “Towards lightweight URL-based phishing detection,” Future Internet, vol. 13, no. 6, p. 154, 2021, doi: 10.3390/fi13060154.
[3] D. Arp, E. Quiring, F. Pendlebury, A. Warnecke, F. Pierazzi, C. Wressnegger, L. Cavallaro, and K. Rieck, “Dos and don’ts of machine learning in computer security,” in Proc. 31st USENIX Security Symp., Boston, MA, USA, Aug. 2022.
[4] R. Zieni, L. Massari, and M. C. Calzarossa, “Phishing or not phishing? A survey on the detection of phishing websites,” IEEE Access, vol. 11, pp. 18499–18519, 2023, doi: 10.1109/ACCESS.2023.3247135.
[5] A. Ejaz, A. N. Mian, and S. Manzoor, “Life-long phishing attack detection using continual learning,” Sci. Rep., vol. 13, Art. no. 11488, 2023, doi: 10.1038/s41598-023-37552-9.
[6] G. Apruzzese, M. Conti, and Y. Yuan, “SpacePhish: The evasion-space of adversarial attacks against phishing website detectors using machine learning,” in Proc. Annu. Comput. Secur. Appl. Conf. (ACSAC), 2022, doi: 10.1145/3564625.3567980.
[7] A. Abuadbba et al., “Towards web phishing detection limitations and mitigation,” arXiv:2204.00985, 2022, doi: 10.48550/arXiv.2204.00985.
[8] M. Sánchez-Paniagua, E. Fidalgo, E. Alegre, W. Al-Nabki, and V. González-Castro, “Phishing URL detection: A real-case scenario through login URLs,” IEEE Access, vol. 10, pp. 42949–42960, 2022, doi: 10.1109/ACCESS.2022.3168681.
[9] O. N. Jasim, “HEURISTIC-PHISH: A lightweight feature-based framework for malicious URL detection,” Muthanna J. Eng. Technol., vol. 14, no. 3, pp. 118–138, 2026, doi: 10.52113/3/eng/mjet/2026-14-03-/117-138.
[10] A. Hannousse and S. Yahiouche, “Towards benchmark datasets for machine learning based website phishing detection: An experimental study,” Eng. Appl. Artif. Intell., vol. 104, Art. no. 104347, 2021, doi: 10.1016/j.engappai.2021.104347.
[11] A. Hannousse, “Web page phishing detection,” Mendeley Data, 2020, doi: 10.17632/c2gw7fy2j4.
[12] S. Das Guptta, K. T. Shahriar, H. Alqahtani, D. Alsalman, and I. H. Sarker, “Modeling hybrid feature-based phishing websites detection using machine learning techniques,” Ann. Data Sci., vol. 11, pp. 217–242, 2022, doi: 10.1007/s40745-022-00379-8.
[13] A. Kulkarni, V. Balachandran, D. M. Divakaran, and T. Das, “From ML to LLM: Evaluating the robustness of phishing web page detection models against adversarial attacks,” Digit. Threats: Res. Pract., vol. 6, pp. 1–25, 2025, doi: 10.1145/3737295.
[14] Y. Yuan, G. Apruzzese, and M. Conti, “Beyond the west: Revealing and bridging the gap between Western and Chinese phishing website detection,” Comput. Secur., vol. 148, Art. no. 104115, 2024, doi: 10.1016/j.cose.2024.104115.
[15] C. Nadeau and Y. Bengio, “Inference for the generalization error,” Mach. Learn., vol. 52, no. 3, pp. 239–281, 2003, doi: 10.1023/A:1024068626366.

📋 How to Cite This Paper

Sharv Magdum, Aditya Jadhav, Aniket Khandagale, Harshal Pawar (2026). Machine Learning versus Simple Rule-Based Baselines for Phishing Website Detection: Effects of Domain-Grouped Evaluation and Feature Sources. International Journal of Computer Techniques, 13(6), 819–823. ISSN: 2394-2231. DOI: https://doi.org/10.5281/zenodo.24348113
© 2026 International Journal of Computer Techniques (IJCT). All rights reserved. · ijctjournal.org
Submit Your Paper