TY  - JOUR
AV  - none
SP  - 154
TI  - Rotation Forest-Based Logistic Model Tree for Website Phishing Detection
N1  - cited By 5; Conference of 21st International Conference on Computational Science and Its Applications, ICCSA 2021 ; Conference Date: 13 September 2021 Through 16 September 2021; Conference Code:265069
PB  - Springer Science and Business Media Deutschland GmbH
SN  - 03029743
EP  - 169
ID  - scholars15658
KW  - Computer crime; Forestry; Machine learning; Personal computing; Websites
KW  -  Cyber security; Logistic model tree; Logistics model; Model trees; Phishing; Phishing attacks; Phishing detections; Rotation forests; Tree method; Web technologies
KW  -  Cybersecurity
N2  - The emergence of web and internet technology has led to its use in a broad array of services ranging from financial to educational services. This has led to a spike in the number of cybersecurity problems over the years, the most notable of which is the phishing attack, in which malicious websites imitate legitimate websites to capture gullible usersâ?? details needed for unauthorized access. However, current mitigation strategies, such as anti-phishing applications and Machine Learning (ML) methods, have been effective for detecting phishing activities. Hackers, on the other hand, are developing new ways to circumvent these countermeasures. Nevertheless, given the dynamism of phishing attempts, there is a continual demand for innovative and efficient solutions for website phishing detection. This study proposes a Rotation Forest-based Logistic Model Trees (RF-LMT) for website phishing detection. LMT is a technique that combines logistic regression and tree inference into a single model tree. Three datasets of different instance distributions, both balanced and imbalanced, are used to investigate the proposed RF-LMT. From the results, it was observed that LMT performed better than the selected baseline classifiers. This finding revealed that LMT can perform comparably to baseline classifiers. However, in comparison to LMT and experimented baseline classifiers, the proposed RF-LMT method showed superior performance in website phishing detection. Specifically, RF-LMT had a high detection accuracy (98.24), AUC (0.998), f-measure (0.982) values with a low false-positive rate (0.018). Furthermore, RF-LMT outperformed existing ML-based phishing attack models. As a result, the proposed RF-LMT method is recommended for dealing with complex phishing attacks. © 2021, Springer Nature Switzerland AG.
Y1  - 2021///
VL  - 12957 
A1  - Balogun, A.O.
A1  - Akande, N.O.
A1  - Usman-Hamza, F.E.
A1  - Adeyemo, V.E.
A1  - Mabayoje, M.A.
A1  - Ameen, A.O.
JF  - Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
UR  - https://www.scopus.com/inward/record.uri?eid=2-s2.0-85115734235&doi=10.1007%2f978-3-030-87013-3_12&partnerID=40&md5=3bec63fef36cf68f8aafb95539bd5f23
ER  -