TY  - JOUR
VL  - 1347
JF  - Communications in Computer and Information Science
A1  - Adeyemo, V.E.
A1  - Balogun, A.O.
A1  - Mojeed, H.A.
A1  - Akande, N.O.
A1  - Adewole, K.S.
N1  - cited By 21; Conference of 2nd International Conference on Advances in Cyber Security, ACeS 2020 ; Conference Date: 8 December 2020 Through 9 December 2020; Conference Code:254989
UR  - https://www.scopus.com/inward/record.uri?eid=2-s2.0-85101586047&doi=10.1007%2f978-981-33-6835-4_41&partnerID=40&md5=b06a500d8bab2d65b3048ed8c1f2b491
Y1  - 2021///
AV  - none
PB  - Springer Science and Business Media Deutschland GmbH
SP  - 627
EP  - 641
ID  - scholars15848
TI  - Ensemble-Based Logistic Model Trees for Website Phishing Detection
KW  - Forestry; Logistic regression; Security of data; Websites
KW  -  Detection accuracy; Dynamic process; Dynamic property; False alarm rate; Logistic models; Phishing attacks; Phishing detections; Phishing websites
KW  -  Computer crime
SN  - 18650929
N2  - The adverse effects of website phishing attacks are often damaging and dangerous as the information gathered from unsuspecting users are used inappropriately and recklessly. Several solutions have been proposed to curb website phishing attacks and to mitigate its impact. However, most of these solutions are rather ineffective due to the evolving and dynamic processes used for phishing attacks. Recently, machine learning (ML)-based solutions are deployed in addressing the phishing attacks due to its ability to deal with the dynamic nature of phishing attacks. Nonetheless, ML solutions suffer drawbacks in the case of high false alarm rates and the need to further improve the detection accuracies of existing ML solutions as proposed in the literature. Considering the dynamism of phishing attacks, there is a continuous need for novel and effective ML-based methods for detecting phishing websites. This study proposed an ensemble-based Logistic Model Trees (LMT) for website phishing attack detection. LMT is the combination of logistic regression and tree induction methods into a single model tree. Experimental results showed that the proposed methods (ABLMT: AdaBoostLMT and BGLMT: BaGgingLMT) are highly effective for website phishing attack detection with the least accuracy of 97.18 and 0.996 AUC values. Besides, the proposed methods outperform some ML-based phishing attack models from recent existing studies. Hence, the proposed methods are recommended for addressing website phishing attacks with dynamic properties. © 2021, Springer Nature Singapore Pte Ltd.
ER  -