tfjs-examples/website-phishing

TensorFlow.js Example: Classify Website URLs as Phishy or Normal

This example shows you how to classify URLs as phishy or normal using Phishing Website Dataset. Since we are classifying the elements of a given set into two groups ie. phishy or normal, this is a binary classification problem.

Features

We have following features available in the dataset:

1. HAVING_IP_ADDRESS: Whether an IP adress is used as an alternate to a domain name. {-1, 1}
2. URL_LENGTH: Whether URL length is legitimate, suspicious or phishing. {1, 0, -1}
3. SHORTINING_SERVICE: Whether it is using URL shortening service or not. {1, -1}
4. HAVING_AT_SYMBOL: Whether URL contains "@" symbol. {-1, 1}
5. DOUBLE_SLASH_REDIRECTING: Whether URL contains double slash redirecting or not. {-1, 1}
6. PREFIX_SUFFIX: Whether URL contains prefix or suffix separated by "-". {-1, 1}
7. HAVING_SUB_DOMAIN: Whether count of sub domains in URL is legitimate, suspicious or phishing. {-1, 0, 1}
8. SSLFINAL_STATE: Whether URL use https and issuer is trusted, use https but issuer is not trusted or no https. {-1, 0, 1}
9. DOMAIN_REGISTERATION_LENGTH: Whether domain expires in less than a year or not. {-1, 1}
10. FAVICON: Whether favicon is loaded from external domain or not. {-1, 1}
11. PORT: Whether port is of preferred status or not. {-1, 1}
12. HTTPS_TOKEN: Whether https token is part of domain or not. {-1, 1}
13. REQUEST_URL: Whether percentage of requests made to external domain falls in legitimate or suspicious category. {-1, 1}
14. URL_OF_ANCHOR: Whether percentage of url in anchor tags reference external domain or self falls in legitimate, suspicious or phishy category. {-1, 0, 1}
15. LINKS_IN_TAGS: Whether percentage of links in meta, script, link tags referencing external domain falls in legitimate, suspicious or phishy category. {-1, 0, 1}
16. SFH: Whether server form handler is empty or contains "about: blank", refers to a different domain or is normal. {1, 0, -1}
17. SUBMITTING_TO_EMAIL: Whether the form submits information to email. {-1, 1}
18. ABNORMAL_URL: Whether URL contains host name or not. {-1, 1}
19. REDIRECT: Whether URL redirects less than equal to 1, between 2 and 4 or greater than 4. {0, 1}
20. ON_MOUSEOVER: Whether onMouseOver changes status bar or not. {-1, 1}
21. RIGHTCLICK: Whether right click is disabled or not. {-1, 1}
22. POPUPWIDNOW: Whether pop up window contain text field or not. {-1, 1}
23. IFRAME: Whether page contains iframe tag or not. {-1, 1}
24. AGE_OF_DOMAIN: Whether age of domain is less than 6 months or not. {-1, 1}
25. DNSRECORD: Whether there is DNS record for the domain or not. {-1, 1}
26. WEB_TRAFFIC: Whether website ranking is less than 100,000, greater than 100,000 or is not recognized by Alexa and/or has no web traffic. {-1, 0, 1}
27. PAGE_RANK: Whether page rank is less than 0.2 or not. {-1, 1}
28. GOOGLE_INDEX: Whether web page is indexed by Google or not. {-1, 1}
29. LINKS_POINTING_TO_PAGE: Whether links pointing to the page is equal to 0, between 0 and 2 or greater than 2. {-1, 0, 1}
30. STATISTICAL_REPORT: Host belongs to top phishing IPs or domains or not. {-1, 1}

Usage

Prepare the environment:

$ npm install
# Or
$ yarn

To build and watch the example, run:

$ yarn watch