Law and Tech Research

Ongoing (since 2024)

Free the Law

Description

Most individuals are likely to face a legal issue at some point in their lives, including problems related to housing, debt, consumer issues and employment. However, their lack of understanding of how to navigate these complex issues prevents them from resolving these issues, causing significant costs to the individuals and society. In a variety of activities, including several PhD studies, we explore new avenues of using technology and AI to increase access to justice, e.g. by bridging the legal literacy gap through the development of automated legal aid systems. This research aims to help people resolve their legal problems and interact with the legal system, thus increasing well-being in society and trust in legal institutions.  

Activities include the: 

• Development of several expert-annotated datasets (e.g., the Belgian Statutory Article Retrieval  Dataset (BSARD) for studying retrieval models aimed at returning the relevant legislation to a short legal question posed in layman’s terms; the Long-form Legal Question Answering (LLeQA) dataset to investigate models aimed at generating long-form responses to any statutory law questions in an interpretable way). 
• Exploration of hybrid search as a strategy to improve in-domain retrieval performance when high-quality domain-specific data is unavailable for training.  
• Testing of a graph-augmented dense statute retriever model that leverages the topological structure of legislation to enrich the article content information and enhance retrieval performance. 
• Leveraging of AI to improve fact descriptions by users to improve legal information retrieval. 
• Automated summarization of court decisions and improved evaluation thereof. 
• Exploration and evaluation of the use of large language model combined with legal expertise to perform legal tasks, e.g. to provide legal information and fill in forms. 
• Use of generative AI to resolve disputes in novel ways. 

 Researchers

Antoine Louis, Semyak Sheth, Lyra Hoeben-Kuil, Hannes Westermann, Jerry Spanakis, Gijs van Dijck, Mindy Nunez Duffourc

More Information

• Interpretable Long-Form Legal Question Answering with Retrieval-Augmented Large Language Models 
• Statutory Article Retrieval via Graph Neural Networks 
• Statutory Article Retrieval Dataset in French (BSARD) 
• Semi-Structured Legal Reasoning and Drafting with Large Language Models 
• Legal aid and AI help poor Americans close 'justice gap' 

Ongoing (since 2024)

RegTech4AI 

Description

The General Data Protection Regulation (GDPR) is a cornerstone of the regulation of AI in the EU. It seeks to facilitate the flow of data across the EU while protecting citizens’ fundamental rights – including data protection and privacy. Even though the GDPR is now more than 7 years old, there remain significant gaps between the law and practice. For example, past research by the project lead showed that less than 10% of mobile apps fulfil the minimum legal requirements regarding consent – one of the core principles of GDPR. As the EU is introducing the AI Act (i.e. its first law aimed directly at AI), it is likely that again – like with GDPR – enforcement will be lagging and that businesses will be overwhelmed by the legal obligations. In response, RegTech4AI will research regulatory technologies (RegTech) to assist enforcement agencies and businesses with AI regulation, and thereby bolster trust in AI systems among citizens. 

 Researchers

Konrad Kollnig, Gijs van Dijck, Qian Li, Kamil Szostak, Bram Rijsbosch, Lucas Giovanni Uberti-Bona Marin

Ongoing (since 2020)

GreaseX: Tackling Digital Harms 

Description 

Harms in user interfaces are widespread. In current scholarship, some of these harms are often referred to as dark patterns or deceptive practices. Other harms, such as hate speech or age-inappropriate content, are also frequently mentioned. However, such categorisations can be too simplistic: what is perceived as harmful often depends greatly on the individual. This is why, in this project, we are exploring how to use state-of-the-art machine learning methods to modify user interfaces on desktop and mobile devices in real-time, and to build personalised models to reduce digital harms. In other words, rather than relying on platforms (and Elon Musk) to take the lead on content moderation, we have been working to devolve some of the power and responsibility for content moderation to end-users. 

 Researchers 

Siddhartha Datta (University of Oxford), Konrad Kollnig  

Ongoing (since 2022)

Lawnotation: Platform for Labeling Legal Data 

Description 

The LAWNOTATION project aims to develop an infrastructure that allows researchers to making legal data and annotation schemes (current and future) accessible for annotation and analysis purposes, to develop an annotation platform for analyzing the linguistic and legal characteristics of legal documents, and to build a user-friendly interface. 
 
A team of developers will work closely together with researchers on the improved access to legal materials. LAWNOTATION is an initiative of the Digital Legal Studies cluster in the Sectorplan Social Sciences and Humanities (SSH) - Rechtsgeleerdheid and other Dutch universities that are collaboratively working on questions related to the digitalisation of law. The research is made possible by the Platform Digitale Infrastructuur SSH 

 Researchers 

prof. Gijs van Dijck (Maastricht University), Hannes Westermann (BISS), Shashank Chakravarthy (BISS), Carlos Aguilera (BISS), prof. Martijn Wieling, prof. Michel Vols (Groningen University), prof. Tom van Engers, prof. Guiseppe Dari-Mattiacci, dr. Balasz Bodó (University of Amsterdam), prof. Andre Janssen, prof. Pietro Ortolani, dr. Pieter Wolters (Radboud University), prof. Floris Bex (Tilburg University / Utrecht University), Robert van Doesburg (TNO / University of Amsterdam). 

Ongoing (since 2019)

Case Law Explorer: Finding Precedents in Case Law 

Description

Traditionally, court decisions are manually read and analyzed in legal education and legal research. However, network analysis based on court citations have shown the potential to answer questions such as: What are the landmark cases in a corpus of decisions?, Which clusters of decisions can be distinguished, and Do certain courts cite different decisions? This project aims at bridging the gap between available computer science techniques for law, and the non-technical legal community. It will provide legal researchers and law students a software platform that visualizes large sets of court decisions as a network, where nodes represent cases and edges represent citations. Users can navigate through the corpus, spot clusters of cases, and access the full case documentation of any selected case in the network. 

This project will provide legal researchers, students, and practitioners an accessible gateway to available technologies applicable to their studies, research or work. This project was funded by a Comenius Teaching Fellows 2019 grant and by a Surf Open Leermaterialen (Web of Law) 2020 grant. 

 Researchers 

Gijs van Dijck, Shashank Chakravarthy, Carlos Aguilera (former researchers: Kody Moodley, Marcel Schaper, Michel Dumontier, Shashank Chakravarthy, Maxine Hanrieder, Bogdan Covrig, Turgay Saba, Pedro V. Hernández Serrano). 

Ongoing (since 2024)

Waste-free, Automated, Legally certain, Legitimate - Euro (WALL-E) transportation and logistics 

Description

The “WALL-E” transportation and logistics research agenda focuses on multidisciplinary and comparative studies addressing multiparty legal structures in transport chains and societal preferences for sustainable technological advancements. It also aims to unite experts from various fields, including law, social sciences, economics, business administration, logistics, and computer science, to explore complex societal issues related to the development of intelligent transport systems. 

Researchers 

Marta Kolacz 

Ongoing (since 2020)

CyberCrimeLinker: Online Criminals as Authors 

Description 

This research addresses the challenge of connecting vendors involved in illegal activities on online markets, a critical task for law enforcement agencies. The manual investigation becomes impractical and resource-intensive with the vast number of vendors and the frequent use of multiple aliases or migration between markets. By leveraging machine learning and authorship attribution techniques, we aim to identify and link these vendors across different platforms, including darknet markets for illegal vendors and escort markets for potential human trafficking vendors. This provides a scalable and effective solution for uncovering hidden connections and aiding in more efficient resource allocation for law enforcement. Additionally, we establish the groundwork to promote ethical authorship attribution by proposing responsible guidelines, ensuring that these approaches are used ethically and responsibly. 

VendorLink

IDTraffickers Dataset

 Researchers 

Vageesh Saxena, Gerasimos Spanakis, Gijs Van Dijck 

Ongoing (since 2023)

Global Privacy Control for Mobile 

Description

Apps and their integrated third-party libraries often collect large amounts of data from Android users to show personalized ads. This can be highly invasive and problematic. Privacy law in California (as of 2021) and Colorado (as of 2024) requires apps and websites to respect opt-outs from ad tracking via the Global Privacy Control (GPC). However, this standard has not yet been widely transposed into mobile apps, and consumer cannot make use of their legally mandated privacy protections. This is why we’ve been studying the level of compliance with the GPC, building the necessary interdisciplinary methodology to do so, and creating reference implementations of GPC for Android. We closely work with responsible regulators in this research. 

 Researchers 

Konrad Kollnig, Sebastian Zimmeck (Wesleyan University)