Making evidence inspectable.

Why Some Judgments Matter More Than Others: Advancing Network Centrality Analysis

Which court decisions are central to European human-rights law, and how much the answer depends on how you measure it.

Citation counts flatten where a judgment sits in the wider network of case law. The paper tests when composite centrality measures do better than their parts, across 66 subnetworks of European Court of Human Rights citations, against two different definitions of importance.

Degree plus eigenvector centrality beat its components in 45 of 66 networks against one ground truth and 43 against the other. The best weighting moved from 0.30 to 0.35 between them, which is the point: what counts as central depends on what you decide importance means.

Prioritising lower-relevance cases helped in 1 of 18 comparisons; prioritising higher-relevance cases helped in 17.

Centrality does not produce an objective ranking of legal importance. The result depends on how the network is built, which ground truth is used, which measures are combined and where the threshold sits. The contribution is an inspectable account of how those choices move judgments from the edge to the centre.

The study →Reproduction repository ↗

Case Law Explorer

A platform that brings Dutch, ECHR and EU court decisions into one schema, so researchers can search, cite and measure them.

Case Law Explorer is a legal-data platform. Dutch (Rechtspraak), European Court of Human Rights (HUDOC) and EU (CELLAR) decisions come in through corpus-specific extraction paths and meet in one schema. Identity, access, product and operations are separate services, so several research tools can share the corpus without sharing one application.

The search interface, shared interface packages, the account service, scoped API access and rate limits, the modular split, Docker deployment and server operations, and parts of the extraction pipeline. The platform builds on earlier Case Law Explorer work and a wider research group; none of this implies sole authorship of the data or the research programme.

• One shared case table keyed by ECLI, with a satellite table per court so each source keeps its own fields instead of being flattened.
• Search and full text cost different amounts, so full judgments load only when a reader asks for them.
• Identity is not entitlement. Signing in says who you are; a separate service decides which endpoints you may use, at what cost, under which limits, and records why.
• A citation graph that was cut short says so in the response, so a truncated network can’t be fed to a centrality measure as if it were complete.

How it is built →app.caselawexplorer.tech ↗

Law & Tech Lab

Law and computer-science methods for research on digital transformation and legal technology: citation-network analysis and research tools for European case law.

Two Peoples, One Land

An atlas of the South Caucasus where every claim is tied to a source, and nothing counts as verified until a person has checked it.

In a contested conflict, a fact, a legal ruling, a state narrative and a historical reading can describe the same event while resting on different institutions and standards of proof. Ordinary narrative pages hide that structure. This prototype tries to make it visible.

AI-assisted drafting is treated as provisional. A claim only becomes verified when a person has checked the source, the locator, the wording and the context. Corrected, disputed, rejected and unresolved claims stay visible, because disagreement is part of the record.

The current pilot covers the Lachin corridor blockade, the September 2023 displacement from Nagorno-Karabakh and the international response. It is a working prototype, not a Maastricht University project, not legal fact-finding, and not a verified scholarly resource unless a record is marked as checked.

How it is built →armenia.wicker.life ↗