An Evaluation of Machine Learning-Based Methods for Detection of Phishing Sites

Miyamoto, Daisuke; Hazeyama, Hiroaki; Kadobayashi, Youki

doi:10.1007/978-3-642-02490-0_66

Cited by 64 publications

(41 citation statements)

References 5 publications

Supporting

Mentioning

Contrasting

Unclassified

Order By: Relevance

“…We have two reasons of employing AdaBoost. One is that it had performed better in our previous comparative study [5], where it demonstrated the lowest error rate, the highest f 1 measure, and the highest AUC of the AdaBoost-based detection method, as mentioned in Section 2. The other is that we expect AdaBoost to cover each user's weak points.Assuming that a user's trust decision can be treated as a classifier, AdaBoost would cover users' weak points by assigning high weights to heuristics that can correctly judge a site that the user is likely to misjudge.…”

Section: Theoretical Backgroundmentioning

confidence: 88%

See 1 more Smart Citation

Behind HumanBoost: Analysis of Users’ Trust Decision Patterns for Identifying Fraudulent Websites

Miyamoto¹,

Hazeyama²,

Kadobayashi³

et al. 2012

JILSA

Self Cite

View full text Add to dashboard Cite

This paper analyzes users' trust decision patterns for detecting phishing sites. Our previous work proposed HumanBoost [1] which improves the accuracy of detecting phishing sites by using users' Past Trust Decisions (PTDs). Web users are generally required to make trust decisions whenever their personal information is requested by a website. HumanBoostassumed that a database of Web user's PTD would be transformed into a binary vector, representing phishing or not-phishing, and the binary vector can be used for detecting phishing sites, similar to the existing heuristics. Here, this paper explores the types of the users whose PTDs are useful by running a subject experiment, where 309 participantsbrowsed 40 websites, judged whether the site appeared to be a phishing site, and described the criterion while assessing the credibility of the site. Based on the result of the experiment, this paper classifies the participants into eight groups by clustering approach and evaluates the detection accuracy for each group. It then clarifies the types of the users who can make suitable trust decisions for HumanBoost.

show abstract

Section: Theoretical Backgroundmentioning

confidence: 88%

“…Our previous work [5] employed nine machine learning techniques for detecting phishing sites. By employing eight heuristics presented by CANTINA, we analyzed 3000 URLs, consisting of 1500 legitimate sites and the same number of phishing sites, reported on PhishTank.com [6] from November 2007 to February 2008.…”

Section: Detection Methods For Phishing Sitesmentioning

confidence: 99%

Behind HumanBoost: Analysis of Users’ Trust Decision Patterns for Identifying Fraudulent Websites

Miyamoto¹,

Hazeyama²,

Kadobayashi³

et al. 2012

JILSA

Self Cite

View full text Add to dashboard Cite

show abstract

“…In contrast to the blacklist method, a heuristic based solution can recognize freshly created phishing websites in real time (Miyamoto, Hazeyama and Kadobayashi 2008). The effectiveness of the heuristic based methods, sometimes called features-based methods, depends on picking a set of discriminative features that could help in distinguishing the type of website (Guang, Jason, et al 2011).…”

Section: Technical Solutionmentioning

confidence: 99%

Tutorial and critical analysis of phishing websites methods

Mohammad

Thabtah

McCluskey

2015

Computer Science Review

132

View full text Add to dashboard Cite

The Internet has become an essential component of our everyday social and financial activities.Internet is not important for individual users only but also for organizations, because organizations that offer online trading can achieve a competitive edge by serving worldwide clients. Internet facilitates reaching customers all over the globe without any market place restrictions and with effective use of e-commerce. As a result, the number of customers who rely on the Internet to perform procurements is increasing dramatically. Hundreds of millions of dollars are transferred through the Internet every day. This amount of money was tempting the fraudsters to carry out their fraudulent operations. Hence, Internet users may be vulnerable to different types of web threats, which may cause financial damages, identity theft, loss of private information, brand reputation damage and loss of customers' confidence in e-commerce and online banking. Therefore, suitability of the Internet for commercial transactions becomes doubtful. Phishing is considered a form of web threats that is defined as the art of impersonating a website of an honest enterprise aiming to obtain user's confidential credentials such as usernames, passwords and social security numbers. In this article, the phishing phenomena will be discussed in detail. In addition, we present a survey of the state of the art research on such attack. Moreover, we aim to recognize the up-to-date developments in phishing and its precautionary measures and provide a comprehensive study and evaluation of these researches to realize the gap that is still predominating in this area. This research will mostly focus on the web based phishing detection methods rather than email based detection methods. INTRODUCTIONAlthough phishing is a relatively new web-threat, it has a massive impact on the commercial and online transaction sectors. Presumably, phishing websites have high visual similarities to the legitimate ones in an attempt to defraud the honest people. Social engineering and technical tricks are commonly combined together in order to start a phishing attack. Typically, a phishing attack starts by sending an e-mail that seems authentic to potential victims urging them to update or validate their information by following a URL link within the e-mail. Predicting and stopping phishing attack is a critical step toward protecting online transactions. Several approaches were proposed to mitigate these attacks. Nonetheless, phishing websites are expected to be more sophisticated in the future. Therefore, a promising solution that must be improved constantly is needed to keep pace with this continuous evolution. Anti-phishing measures may take several forms including: legal, education and technical solutions. To date, there is no complete solution able to capture every phishing attack. The Internet community has put in a considerable amount of effort into defensive techniques against phishing. However, the problem is continuously evolving and ever more complicated decept...

show abstract

“…In 2010, a survey presented in [7] aimed to evaluate the performance of machine-learning-based-detection-methods including: "AdaBoost, Bagging, SVM, Classification and Regression Trees, Logistic Regression, Random Forests, NN, Naive Bayes and Bayesian Additive Regression Trees" showed that 7 out of 9 of machine-learning-based-detection-methods outperformed CANTINA in predicting phishing websites those are: "AdaBoost, Bagging, Logistic Regression, Random Forests, Neural Networks, Naive Bayes and Bayesian Additive Regression Trees". A dataset consisting of 1500 phishing websites and 1500 legitimate websites used in the experiments.…”

Section: Tf-idf (Term Frequency-inverse Document Frequency)mentioning

confidence: 99%

“…In contrast to the blacklist method, a heuristic-based solution can recognize freshly created phishing websites in real-time [7]. The effectiveness of the heuristic-based methods, sometimes called features-based methods, depends on picking a set of discriminative features that could help in distinguishing the type of website [8].…”

Section: Introductionmentioning

confidence: 99%

Predicting phishing websites based on self-structuring neural network

Mohammad

Thabtah

McCluskey

2013

Neural Comput & Applic

271

131

View full text Add to dashboard Cite

-Internet has become an essential component of our everyday social and financial activities. Nevertheless, internet-users may be vulnerable to different types of web-threats which may cause financial damages, identity theft, loss of private information, brand reputation damage and loss of customer's confidence in ecommerce and online banking. Phishing is considered as a form of web-threats that is defined as the art of impersonating a website of an honest enterprise aiming to obtain confidential information such as usernames, passwords and social security number. So far, there is no single solution that can capture every phishing attack. In this article, we proposed an intelligent model for predicting phishing attacks based on Artificial Neural Network "ANN" particularly self-structuring neural networks. Phishing is a continuous problem where features significant in determining the type of webpages are constantly changing. Thus, we need to constantly improve the network structure in order to cope with these changes. Our model solves this problem by automating the process of structuring the network and shows high acceptance for noisy data, fault tolerance and high prediction accuracy. Several experiments were conducted in our research, the number of epochs differs in each experiment. From the results, we find that all produced structures have high generalization ability.

show abstract

An Evaluation of Machine Learning-Based Methods for Detection of Phishing Sites

Cited by 64 publications

References 5 publications

Behind HumanBoost: Analysis of Users’ Trust Decision Patterns for Identifying Fraudulent Websites

Behind HumanBoost: Analysis of Users’ Trust Decision Patterns for Identifying Fraudulent Websites

Tutorial and critical analysis of phishing websites methods

Predicting phishing websites based on self-structuring neural network

Contact Info

Product

Resources

About