The dataset contains extracted attributes from websites that can be used for Classification of webpages as malicious or benign. The dataset also includes raw page content including JavaScript code that can be used as unstructured data in Deep Learning or for extracting further attributes. The data has been collected by crawling the Internet using MalCrawler [1]. The labels have been verified using the Google Safe Browsing API [2]. Attributes have been selected based on their relevance [3]. The details of dataset attributes is as given below: 'url' - The URL of the webpage. 'ip_add' - IP Address of the webpage. 'geo_loc' - The geographic location where the webpage is hosted. 'url_len' - The length of URL. 'js_len' - Length of JavaScript code on the webpage. 'js_obf_len - Length of obfuscated JavaScript code. 'tld' - The Top Level Domain of the webpage. 'who_is' - Whether the WHO IS domain information is compete or not. 'https' - Whether the site uses https or http. 'content' - The raw webpage content including JavaScript code. 'label' - The class label for benign or malicious webpage.

Python code for extraction of the above listed dataset attributes is attached. The Visualisation of this dataset and it python code is also attached. This visualisation can be seen online on Kaggle [5].

The description of the attributes from the Dimension class in version 1.0 of the CSD model.

The description of the attributes from the DependentVariable class in version 1.0 of the CSD model.

Context

This dataset has been prepared to carryout classification of webpages as malicious or benign.

Content

The dataset contains extracted attributes from websites that can be used for Classification of webpages as malicious or benign. The dataset also includes raw page content including JavaScript code that can be used as unstructured data in Deep Learning or for extracting further attributes. The data has been collected by crawling the Internet using MalCrawler [1]. The labels have been verified using the Google Safe Browsing API [2]. Attributes have been selected based on their relevance [3].

References

[1] Singh, A. K., and Navneet Goyal. "MalCrawler: A crawler for seeking and crawling malicious websites." In International Conference on Distributed Computing and Internet Technology, pp. 210-223. Springer, Cham, 2017. [2] https://developers.google.com/safe-browsing [3] Singh, A. K., and Navneet Goyal. "A Comparison of Machine Learning Attributes for Detecting Malicious Websites." In 2019 11th International Conference on Communication Systems & Networks (COMSNETS), pp. 352-358. IEEE, 2019.

Inspiration

The dataset seeks to address classification of webpages using machine learning techniques.

Overview of Data

This dataset is a data dump containing data from June 2008 to March 2013. Note that Stack Overflow originated only in June 2008. Therefore, this dump includes all the questions and answers on Stack Overflow until March 2013.

Stack Overflow provides data dumps of all user generated data, including questions asked with the list of answers, the accepted answer per question, up/down votes, favourite counts, post score, comments, and anonymized user reputation. Stack Overflow allows users to tag discussions and has a reputation-based mechanism to rank users based on their active participation and contributions.

Attribute Information

Attribute info the datasets are in xml format including questions and answers for the following topics:

* CSS
* CSS-mobile
* HTML5
* HTML5-mobile
* JavaScript
* Javascript-mobile

The datasets used for this manuscript were derived from multiple sources: Denver Public Health, Esri, Google, and SafeGraph. Any reuse or redistribution of the datasets are subjected to the restrictions of the data providers: Denver Public Health, Esri, Google, and SafeGraph and should consult relevant parties for permissions.1. COVID-19 case dataset were retrieved from Denver Public Health (Link: https://storymaps.arcgis.com/stories/50dbb5e7dfb6495292b71b7d8df56d0a )2. Point of Interests (POIs) data were retrieved from Esri and SafeGraph (Link: https://coronavirus-disasterresponse.hub.arcgis.com/datasets/6c8c635b1ea94001a52bf28179d1e32b/data?selectedAttribute=naics_code) and verified with Google Places Service (Link: https://developers.google.com/maps/documentation/javascript/reference/places-service)3. The activity risk information is accessible from Texas Medical Association (TMA) (Link: https://www.texmed.org/TexasMedicineDetail.aspx?id=54216 )The datasets for risk assessment and mapping are included in a geodatabase. Per SafeGraph data sharing guidelines, raw data cannot be shared publicly. To view the content of the geodatabase, users should have installed ArcGIS Pro 2.7. The geodatabase includes the following:1. POI. Major attributes are locations, name, and daily popularity.2. Denver neighborhood with weekly COVID-19 cases and computed regional risk levels.3. Simulated four travel logs with anchor points provided. Each is a separate point layer.

Mapping of CSD model attribute values to JSON serialized values.