Home // University // News // Latest News // Open and transparent access to legal data

Open and transparent access to legal data

twitter linkedin facebook google+ email this page
Published on Wednesday, 31 May 2017

The Ministry of State’s Service central de législation (SCL) has partnered up with the Interdisciplinary Centre for Security, Reliability and Trust (SnT) at the University of Luxembourg to develop advanced legal text processing technologies that will dramatically improve how citizens, and legal experts alike, will use and interact with laws and regulations.

The collaborative project between SCL and SnT, also backed by Digital Lëtzebuerg, has already begun to bear fruit. Earlier in 2017, SCL gave a major boost to “LegiLux” - a digital legal portal that provides access to legal information for Luxembourg’s citizens including the Grand Duchy’s Official Journal. The collaboration between SCL and SnT played a key role in processing the existing legal Codes and converting them into the web-enabled and machine-analysable format used by the new LegiLux. LegiLux visitors range from the general public, to administrative personnel and more specific audiences within the legal profession. Since its launch in January 2017, the new LegiLux website has been accessed over 12 million times, with over 13 million searches performed.

Modernizing legal documents

In the past, laws and regulations were written exclusively with the intention of being read by humans, and not being analysed by machines. Traditionally, Legilux, like other national legal portals around the world, had relied on print-only formats such as PDF for the publication and dissemination of legal documents. “It has long been known that print-only formats are quite limited when it comes to building smart, interactive legal services. We need flexible formats that support not only the traditional way of working with legal texts but also more advanced and computer-assisted forms of interaction,” says John Dann, the director of SCL. “New technologies can provide added value by automatically deriving useful insights and analytics from legal documents. Connecting different sources of information like legislation and statistics or GPS information has now become an attainable goal. To benefit from such technologies, we have to focus on machine-analysable formats with embedded explicit information about the structure and semantics of legal content,” he continues.

A key question that comes up is what to do with the thousands of pages of existing legal Codes. “Manually transforming the legal corpus for an entire country from plain text into a digital, machine-readable format is no easy feat, especially considering all the cross-references in legal articles that link different provisions. Put simply, doing this task manually while maintaining quality is prohibitively expensive and time consuming,” states Dr. Mike Sabetzadeh, the lead researcher on the project.

Legilux, Journal officiel du Grand-Duché de Luxembourg

 

Therefore, a team of research scientists, including Dr. Nicolas Sannier, Dr. Mike Sabetzadeh, and Prof. Lionel Briand, has been developing a tool named ARMLET (Automated Retrieval of Metadata from Legal Texts). Using sophisticated natural language processing, ARMLET automatically detects the structural elements of legal texts (for example, chapters, articles, paragraphs) and creates a document with navigable links for the cross-references. “We already benefitted from ARMLET during the rollout of the new LegiLux website. As an example, the Civil Code is now available in a modern layout with navigable cross-references. With the ARMLET solution, 90% of the structuring work – a time consuming and repetitive task – was already taken care of. Over time, our goal is to apply the tool to transform most of the legal Codes in Luxembourg,” says J. Dann.

Luxembourg has long been a pioneer of web-enabled formats of legal texts. In particular, J. Dann is the founder of and the key driving force behind the European Legislation Identifier (ELI). “ELI is an intuitive naming framework for the semantic elements of legal texts, cross-references being an example of these elements. By design, ELI is independent of countries and legal jurisdictions. Ultimately, ELI will allow us to harmonise the representation of legislative texts within the EU, so that that these texts can be seamlessly linked,” says J. Dann.

“With the help of John [Dann] and his team at SCL, we would like to broaden our expertise and propose the solutions we have developed to other countries. With this goal in mind, we are now working on a tailorable technology that can automatically and accurately transform legacy legal texts into a modern, digital format,” states Prof. Briand, Vice-Director of SnT and Head of the Software Verification and Validation Lab (SVV), the SnT research unit collaborating with SCL.

Much more on the horizon

As a next step, SCL and SnT are now investigating how their tools can be applied to the field of legal compliance. ”A major source of difficulty here is that the requirements that must be met for compliance are often not formulated in a machine-interpretable way. We are employing natural language processing and machine learning techniques to automatically extract compliance requirements from legal texts, and transform these requirements into precise rules,” Dr. Sabetzadeh says.

“In addition to the generous long-term financial support from Fonds National de la Recherche (FNR), the Software Verification and Validation Laboratory benefits greatly from our close collaboration with industry and public-service partners in all of our projects. The results of our collaboration with SCL is an excellent example of what we aspire to achieve via research. What we did so far is only the beginning of what promises to be a long and fruitful research activity,” Prof. Briand says.