MAXIMUM PHISH BAIT: TOWARDS FEATURE BASED DETECTION OF PHISING USING MAXIMUM ENTROPY CLASSIFICATION TECHNIQUE

Asani, A and Adegun, A and Adekanmi, A and Oluwatobi, E MAXIMUM PHISH BAIT: TOWARDS FEATURE BASED DETECTION OF PHISING USING MAXIMUM ENTROPY CLASSIFICATION TECHNIQUE. MAXIMUM PHISH BAIT: TOWARDS FEATURE BASED DETECTION OF PHISING USING MAXIMUM ENTROPY CLASSIFICATION TECHNIQUE. In: iSTEAMS Research Nexus Conference 2014, 29th-31st May, 2014, ABUAD, Nigeria..

[img]
Preview
Text
Feature%20based%20Phishing%20detectioin%20using%20Maximum%20Entropy%20Classification%20Technique.pdf

Download (1MB) | Preview
[img] Other (Plain Text Bibliography)
bibliography.txt - Bibliography

Download (3kB)

Abstract

Several antiphishing methods have been employed with the primary task of automatically apprehending and ruling out or preventing phishing e-mail from users’ mail stream. Phishing attacks pose great threat to internet users and the extent can be enormous if unchecked. Two major category techniques that have been shown to be useful for classifying e-mail messages automatically include the rule based method which classifies email by using a set of heuristic rules and the statistical based approach which model e-mails statistically usually under a machine learning framework. The statistical based methods have been found in literature to outperform the rule based method. This study proposes the use of the Maximum Entropy Model, a generative model and show how it can be used in anti-phishing tasks. The model based feature proposed by Bergholz et al (2008) will also be adopted. This has been found to outperform basic features proposed in previous studies. An experimental comparison of our approach with other generative and non-generative classifiers is also proposed. This approach is expected to perform comparably better than others method especially in the elimination of false positives. Keywords: Antiphishing, Rule-based, Statistical-based, Machine learning, Maximum Entropy Model, generative classifiers, non-generative classifier

Item Type: Article
Subjects: A General Works > AI Indexes (General)
Depositing User: Mr Tayo Okunlola
Date Deposited: 02 Jan 2015 04:41
Last Modified: 02 Jan 2015 04:41
URI: http://eprints.abuad.edu.ng/id/eprint/29

Actions (login required)

View Item View Item