Presentation is loading. Please wait.

Presentation is loading. Please wait.

Report : 鄭志欣 Advisor: Hsing-Kuo Pao 1 Learning to Detect Phishing Emails I. Fette, N. Sadeh, and A. Tomasic. Learning to detect phishing emails. In Proceedings.

Similar presentations


Presentation on theme: "Report : 鄭志欣 Advisor: Hsing-Kuo Pao 1 Learning to Detect Phishing Emails I. Fette, N. Sadeh, and A. Tomasic. Learning to detect phishing emails. In Proceedings."— Presentation transcript:

1 Report : 鄭志欣 Advisor: Hsing-Kuo Pao 1 Learning to Detect Phishing s I. Fette, N. Sadeh, and A. Tomasic. Learning to detect phishing s. In Proceedings of the International World Wide Web Conference (WWW), pages 649–656, 2007.

2 Outline 2 Introduction Method Empirical evaluation Conclusion

3 Introduction 3 Phishing (Spoofed websites) Stealing account information Logon credentials Identity information Phishing Problem – Hard

4 Method 4 PILFER – A Machine Learning based approach to classification. phishing s / ham (good) s Feature Set Features as used in classification

5 5 IP-based URLs: Phishing attacks are hosted off of compromised PCs. This feature is binary.

6 6 Age of linked-to domain names Legitimate-sounding domain name Palypal.com paypal-update.com These domains often have a limited life WHOIS query date is within 60 days of the date the was sent – “fresh” domain. This is a binary feature

7 7 Nonmatching URLs This is a case of a link that says paypal.com but actually links to badsite.com. Such a link looks like paypal.com. This is a binary feature.

8 8 “Here” links to non-modal domain “Click here to restore your account access” Link with the text “link”, “click”, or “here” that links to a domain other than this “modal domain” This is a binary feature.

9 9 HTML s s are sent as either plain text, HTML, or a combination of the two - multipart/alternative format. To launch an attack without using HTML is difficult. This is a binary feature.

10 10 Number of links The number of links present in an . in HTML tag This is a continuous feature.

11 11 Number of domains Simply take the domain names previously extracted from all of the links, and simply count the number of distinct domains. Look at the “main” part of a domain https://www.cs.university.edu/ This is a continuous feature.

12 12 Number of dots Subdomains like Redirection script, such as This feature is simply the maximum number of dots (`.') contained in any of the links present in the , and is a continuous feature.

13 13 Contains javascript Attackers can use JavaScript to hide information from the user, and potentially launch sophisticated attacks. An is flagged with the “contains javascript” feature if the string “javascript” appears in the , regardless of whether it is actually in a or tag This is a binary feature.

14 14 Spam-filter output This is a binary feature, using the trained version of SpamAssassin with the default rule weights and threshold. “Ham” or “Spam” This is a Binary feature.

15 Empirical Evaluation 15 Machine-Learning Implementation Testing Spam Assassin Datasets Additional Challenges False Positives vs. False Negatives

16 16 Machine-Learning Implementation-PILFER First, run a set of scripts to extract all the features listed. Second, we train and test a classifier using 10-fold cross validation. Random Forest (classifier) Random forests create a number of decision trees and each decision tree is made by randomly choosing an attribute to split on at each level, and then pruning the tree.

17 17 we use a random forest as a classifier.

18 18 Testing SpamAssassin SpamAssassin is a widely-deployed freely-available spam filter that is highly accurate in classifying spam s. We classify the exact same dataset using SpamAssassin version 3.1.0, using the default thresholds and rules. Using “Untrain” SpamAssassin “Training” on 10-fold

19 19 Datasets Two publicly available datasets. ham corpora from the SpamAssassin project 6950 non-phishing non-spam s Phishingcorpus approximately 860 messages

20 20 Additional Challenges The age of the dataset. Phishing websites are short-lived. Some of our features can therefore not be extracted from older s, making our tests difficult. EX: Domain linked to

21 Result 21

22 22

23 Conclusion 23 it is possible to detect phishing s with high accuracy by using a specialized filter, using features that are more directly applicable to phishing s than those employed by general purpose spam filters.

24 Reference 24 I. Fette, N. Sadeh, and A. Tomasic. Learning to detect phishing s. In Proceedings of the International World Wide Web Conference (WWW), pages 649–656, Phishing%20 s.pptx 0Phishing%20 s.pptx cht.blogspot.com/2010/01/phishing-mail.html cht.blogspot.com/2010/01/phishing-mail.html

25 25


Download ppt "Report : 鄭志欣 Advisor: Hsing-Kuo Pao 1 Learning to Detect Phishing Emails I. Fette, N. Sadeh, and A. Tomasic. Learning to detect phishing emails. In Proceedings."

Similar presentations


Ads by Google