In this line of research, we focus on the online systems and the privacy mechanisms they provide. In an ideal world, users are able to understand online privacy practices and have access to usable privacy settings that control the granularity of the data collected about them. However, existing privacy policies and settings of online applications are often hard to comprehend by an average user, are unreachable, and lack effective controls over the collection and release of user data. As a result, users are unlikely to be aware of privacy risks and are incapable of exercising informed control when interacting with websites, apps, devices, and services. We have attempted to mitigate these privacy risks by creating technological solutions to:
- Make privacy policies more readable (PriBot )
- Analyze the impact of General Data Protection Regulation (GDPR) on the landscape of privacy policies
- Create an automated system to find privacy settings, transform them into machine readable format and present them to users in a user-friendly interface where the users can search for privacy settings of interest.
Note: This research is funded by the NSF under grant #1942014 (CAREER: Presentation and Mitigation of Privacy Risks for Online Users)
In this project, we are lookings at online users' interaction with privacy controls. Online service providers offer online menus and forms so users can control their privacy settings. However, in their current form, privacy control settings suffer from usability and reachability issues, making it hard for users to exercise informed control. To mitigate this and make the privacy controls more usable, we designed and developed PriSEC - a privacy setting enforcer that transforms privacy control pages from domains using machine learning techniques into a machine-readable format. Applications can be built on this machine-readable format to enhance the user experience. We demonstrated this by developing a browser extension that leverages a search interface where the users can easily select and enforce the desired setting.
Surfacing Privacy Settings Using Semantic Matching
Online services utilize privacy settings to provide users with control over their data. However, these privacy settings are often hard to locate, causing the user to rely on provider-chosen default values. In this work, we train privacy settings centric encoders and leverage them to create an interface that allows users to search for privacy settings using free-form queries. To achieve this, we create a custom Semantic Similarity dataset, which consists of real user queries covering various privacy settings. We then use this dataset to fine-tune the state of the art encoders. Using these fine-tuned encoders, we perform semantic matching between the user queries and the privacy settings to retrieve the most relevant setting. Finally, we also use these encoders to generate embeddings of privacy settings from the top 100 websites and perform unsupervised clustering to learn about the online privacy settings types.
In this project, we studied the impact of the General Data Protection Regulation (GDPR) on the landscape of privacy policies online. We conducted the first longitudinal, in-depth, and at-scale assessment of privacy policies before and after the GDPR. We gauged the complete consumption cycle of these policies, from the first user impressions until the compliance assessment. To achieve this, we created a diverse corpus of two sets of 6,278 unique English-language privacy policies from inside and outside the EU, covering their pre-GDPR and post-GDPR versions. We further developed a workflow for the automated assessment of requirements in privacy policies using natural language processing techniques.