Phishing and Legitimate Webpage Features for Machine Learning, 2015-2017

10,000 webpages (5,000 phishing and 5,000 legitimate) were downloaded between January 2015 and June 2017. This dataset contains 48 features extracted from each webpage using an improved technique based on the Selenium WebDriver browser automation framework. The dataset was created by Shashwat Tiwari and is associated with research by Tan, Choon Lin from 2018.

Use Cases

• Benchmarking phishing webpage classification models based on the 48 extracted features.
• Conducting rapid proof-of-concept experiments for feature evaluation in anti-phishing systems.
• Analyzing the effectiveness of browser automation versus regular expression parsing for feature extraction.

Strengths

• Contains a balanced set of 5,000 phishing and 5,000 legitimate webpage instances.
• Features were extracted using a precise and robust method leveraging the Selenium WebDriver framework.
• Clear temporal coverage spanning two collection periods from 2015 to 2017.

Limitations

• Column-level documentation is absent; field semantics must be inferred after download.
• Row count is unknown for the specific file on OpenML, which may limit suitability assessment.
• Last update date is unknown; freshness unverified.

Provenance

Source

Mendeley Data (doi: 10.17632/h3cgnj8hft.1), hosted on Kaggle and OpenML.

Collection Method

Webpages were downloaded and 48 features were extracted using Selenium WebDriver.

Time Range

January to May 2015 and May to June 2017.

Freshness

Data collected up to 2017; last platform update unknown.

Geography

null