NWANAH, CHRISTOPHER CHIDOZIE (2026) AI-POWERED PHISHING WEBSITE DETECTION TOOL USING A TRIBRID AI MODULE. Other thesis, Godfrey Okoye University, Enugu Nigeria.
![[img]](https://eprints.gouni.edu.ng/style/images/fileicons/text.png) |
Text
Nwanah Christopher Project Tribrid AI Chapter 1 - 5 Real work.pdf Download (1MB) |
Abstract
This study explores the development and design of an Artificial Intelligence-based phishing website detector through a Tribrid AI Module. This work aims to solve the current issue of phishing threats, whereby scammers use websites, emails, or links that mimic genuine websites to obtain sensitive data. Some of the existing approaches such as blacklist and rules-based methods face difficulties because of the fact that such approaches require records of attacks and fail in situations where attackers come up with new phishing links. To solve this challenge, the system proposes to combine three different intelligent classifiers, including Random Forest, Support Vector Machine, and Neural Network into a tribrid decision making model. The system analyzes the lexical, host, content, and email-based characteristics of user-provided URLs or email content, analyzes these features through the three classifiers and aggregates the results of the three models using weighted voting technique to produce a phishing score and justification. Python, Flask, Scikit-learn, TensorFlow/Keras, Pandas, NumPy, HTML, CSS, JavaScript and SQLite libraries are the technologies that will be used in this study, alongside the public phishing and non-phishing URL/email datasets. The developed tool consists of the following elements: URL scanning, email scanning, confidence scoring, scan history, user authentication, and user alerts. The performance evaluation of the Tribrid AI Module on the test set revealed 96.8% accuracy, 95.9% precision, 96.1% recall, and 96.0% F1-score. The number of legitimate URLs detected by the system was 3,784, while the number of phishing URLs detected was 2,403. There were 103 false positives and 97 false negatives. The Area Under the ROC Curve (AUC) was 0.977, which proves high discrimination capabilities at different thresholds of classification. The paper offers a practical, interpretable, and extendable anti-phishing solution that can be used by individuals, students, researchers, and businesses.
| Item Type: | Thesis (Other) |
|---|---|
| Subjects: | Q Science > Q Science (General) |
| Divisions: | Faculty of Computing And Information Technology (FACIT) |
| Depositing User: | Cynthia Ugwuoti |
| Date Deposited: | 28 Jul 2026 13:12 |
| Last Modified: | 28 Jul 2026 13:12 |
| URI: | http://eprints.gouni.edu.ng/id/eprint/6062 |
Actions (login required)
 |
View Item |
