Hello, there! Welcome to the Github page of my MSc Cyber Security project. This is a study about Cookie Banners, and the privacy options that they provide, in Greece and the UK. By extending OpenWPM, we were able to scrape over 14k websites and collect 8k Cookie Banners.
In order to crawl that many websites and collect their Cookie Notices, this project developed a set of novel methods and extended existing ones. For instance, in order to identify the most pupolar websites in Greece and the UK, a number of different tools were built to parse the Tranco and verify that the extracted websites allow crawling by checking their robots.txt and Terms of Service pages.
Moreover, OpenWPM, a popular web privacy measurement framework, was extended so that it identifies Cookie Banners with the aid of the I don't care about cookies list.
You can find all the code and how to run it under the code directory of this Git page. Why not give it a go in websites of your country? :)
Our researchers has been accepted to and to appear in the proceedings of IFIP SEC 2021! You can read it by...- Downloading it directly from arXiv
- Pulling it from our GitHub repo
- Building it from source as shown below (Latex is required)
# fork the repo
cd paper
./build.shPlease cite the following paper if you use this repository:
George Kampanos and Siamak F. Shahandashti. "Accept All: The Landscape of Cookie Banners in Greece and the UK". To appear in the proceedings of the 36th International Conference on ICT Systems Security and Privacy Protection (IFIP SEC 2021)
@misc{kampanos2021accept,
title={Accept All: The Landscape of Cookie Banners in Greece and the UK},
author={Georgios Kampanos and Siamak F. Shahandashti},
year={2021},
eprint={2104.05750},
archivePrefix={arXiv},
primaryClass={cs.CR}
}
Our findings show that almost half of the websites we surveyed display a cookie banner. More specifically,around 48% of Greek and 44% of the UK websites included a cookie notice.
The most prevalent number of options in both countries is two. The mean numberof options is 2.1 for Greece and 1.8 for the UK. Worryingly, a considerable proportion of cookie banners provide either no option or only one option to the user. As the figure below shows, althoughAffirmative options are quite ubiquitous in cookie banners in both countries(Greece: 95%, UK: 88%), Negative options are quite rare (Greece: 20%, UK:6%)
To get a more comprehensive view of the connotations relayed by cookiebanner texts in the UK, we performed an automated sentiment analysis of thewords used in all UK banner texts using NRCLex. The analysis found that agenerally positive emotional affect was present in around 80% of the banner texts,whereas a generally negative affect was present in only around 14%. Besides, anoverwhelming majority (of more than 9 in 10) of the texts with a negative affectalso had a positive effect present as well. Looking at more specific emotionalaffects, trust and joy are among the most prevalent, present in around 66% and46% of the texts, respectively. The prevalence of general and specific emotionalaffects is shown in the figure below.
