JuDDGES Open Wednesday Seminar

Data Management Plans (DMPs) are crucial for a structured research data management and often a mandatory part of research proposals. The manual creation of DMPs can be very time-consuming, since many researchers have to start from scratch, are unsure about the required content and may run the risk of not meeting the funder requirements. By using tools, DMPs can be effectively developed and managed. There are a variety of tools to support the development of DMPs: from discipline agnostic DMP tools, which can be used to generate a generic draft DMP, to discipline-specific DMP tools, which support the creation of a DMP in a specific research field, such as psychology, biodiversity, engineering, or the life sciences. Our aim was to develop a quick and easy to use DMP service for members of Saxon research institutions, building on existing work. In order to evaluate 18 of the existing DMP tools, we defined 32 requirement parameters covering aspects with regard to basic functions, technical aspects and user-friendliness. To further prioritize, a weight factor between zero (not relevant) and three (high priority) was assigned to every requirement parameter. The DMP tools were rated according to a fixed rating scheme from zero (poor) to ten (excellent), and then multiplied with the weight factor. Our results can support tool developers to identify potential improvements and hosting institutions to select a tool suited to their specific needs.

De la décision rendue à la décision “donnée” : petite histoire de la diffusion des décisions de justice en France.

Cette présentation reviendra sur les trajectoires de diffusion suivies par les décisions de justice depuis la Révolution Française. Après être revenue sur les premières initiatives de publication, elle se penchera sur la numérisation des modes de diffusion dans les années 1960/1980, puis sur la question de l’open data des décisions de justice. Les modes de publication et de diffusion variant selon l’ordre ayant rendu la décision (administratif ou judiciaire), mais également selon le degré de juridiction (première instance, appel, Cour de Cassation ou Conseil d’État), il s’agira de distinguer ces trajectoires différentes, et d’analyser leurs évolutions (conjointes ou non) au fil du temps.

La loi du 7 octobre 2016 a suscité ou renforcé des projets d’exploitation des décisions de justice, étendus aux décisions de première instance annoncées comme accessibles et gratuites dans un avenir relativement proche. C’est dans ce mouvement qu’ont été développés des outils algorithmiques d’analyse de décisions. Bien avant ce mouvement, le monde de la recherche sur le droit avait développé des analyses de contentieux, restées assez peu nombreuses. Les questions posées sont variables : quels sont les demandeurs et défendeurs, quelle partie obtient gain de cause, quels arguments sont développés, quelles solutions sont retenues… ? Ces travaux produisent des connaissances nouvelles sur l’activité juridictionnelle à partir de l’analyse des (ir)régularités contenues dans les décisions. On peut également envisager que ces régularités soient utilisées pour fournir des outils d’aide à la décision. Dès lors que ces outils prétendent influencer les décisions de justice, leur fiabilité devait être interrogée. Par ailleurs, le recours à ces méthodes dans une perspective de connaissance, séduisant, devait être testé. C’est à ces préoccupations qu’a tenté de répondre le groupe de travail interdisciplinaire « e-Juris » et ce sont ses conclusions, très mitigées, que cet article retrace.

The overall goal of the JuDDGES project is to harness state-of-the-art Natural Language Processing & Human-In-The-Loop technologies to provide legal researchers with new Open software and tools that enable extensive, flexible and on-going meta-annotation capability (both automated and employing domain experts in-the-loop). This capability is applied to legal records/judgments from criminal courts across jurisdictions with varied legal constitutions (Poland, England & Wales).

Presentation here

Data sharing is widely recognised as crucial for accelerating scientific research and enhancing its quality. However, it remains far from a common practice. Funding agencies occupy a key position to facilitate the sharing of research data by both offering incentives and requiring data sharing as part of their policies and grant conditions. Despite this, relatively little is known about their experiences in this area. Drawing on the findings of two interview-based studies, this presentation reports on the challenges faced by international funding agencies in their efforts to promote and implement data sharing through data-related policies.

Presentation here

Riza Batista-Navarro is Senior Lecturer in Text Mining at the University of Manchester. She has been working on the development of natural language processing (NLP) methods for analysing online hate speech, as part of the OSR4Rights project (https://osr4rights.org/) and a collaboration with the Centre for Information Resilience (CIR) researching technology-facilitated gender-based violence in Ethiopia (https://aclanthology.org/2024.rail-1.13/). This presentation will outline key elements of the policy paper/recommendations resulting from their empirical work with CIR.

Legal scholarship in the United States developed differently than legal scholarship in other parts of the world or other scholarship in the United States. These differences have led to the wide availability of journal articles on the internet without paywalls, but without the hallmarks of Open Access publications in other fields. Journals may hold a copyright on these articles, signed over by the author. Very few are licensed for reuse. And it is difficult to know what the policies are because the journals rarely post them publicly. This presentation will discuss the current state of Open Access in US legal journals and the challenges resulting from the lack of clear policies from the journals.

Presentation here

Challenging the narrative: DMPs as best practice – fact or fiction?

For years, Data Management Plans (DMPs) have been championed as a cornerstone of responsible research. Is this reputation based on evidence or primarily on professional consensus? Are their benefits real or simply assumed? Originally designed as bespoke tools for complex projects, DMPs have evolved into mandatory requirements by funders and institutions, promoted as essential for managing data throughout the research lifecycle, demonstrating integrity and enabling sharing and reuse. But do these claims hold up under scrutiny? This presentation examines the origins of DMPs, the rationale behind mandates, and what research (or its absence) reveals about their actual impact. It explores criticisms such as administrative burden, poor quality and “tick-box” solutions and asks whether DMPs genuinely improve practice. Finally, it looks ahead to alternative approaches and emerging technologies, such as AI, that could be employed by existing and new digital research infrastructures to streamline data management and free researchers to focus on what matters most: their research.

Presentation here