Find relevant decisions.
Discovery & retrieval
Combine semantic search with explicit inclusion criteria. Expand through related terminology and inspect the full decisions when excerpts leave context unresolved.
Legal analytics
Court decisions,
made comparable.
Legal judgments contain facts, arguments and outcomes in varied forms. We use semantic analysis to structure that material for systematic research.
Core principleCompare like with like. Keep the original decision within reach.
Interactive demo
Judgment research workspace.
Search fictional examples, adjust filters and inspect the structured records. Everything runs in your browser.
Sample records appear here when JavaScript is enabled.
Demo data only. Results use keyword matching on fictional records, without live research or API calls.
Methods
From a body of judgments
to a defined research question.
The task extends beyond finding a phrase. It requires understanding what the document says, where it says it and whether the result is comparable.
Extract with context.
Structured interpretation
Organize relevant facts, legal issues and outcomes into defined fields. Language models assist with interpretation; deterministic checks test format and consistency.
Compare within groups.
Analysis & review
Group decisions by relevant legal and factual conditions. Account for differences in case composition, and review unusual findings against the original text.
Selected work / methods
Foundations in judgment research.
Selected methods from prior research projects.
Judgment research
Structured decision analysis
Converted public court decisions into comparable research records. Combined semantic extraction with rule-based checks and retained source references for review.
Document discovery
Search and extraction review
Used semantic search and related terms to locate relevant decisions. Reviewed full documents to identify information missed in shorter excerpts.
Research discipline
Differences in outcomes need context. Our comparative work uses defined groups and reviews the effect of case composition. Descriptive patterns are not treated as proof of causation.
Research outputs
Evidence that can be examined.
Outputs are organized around the research question and the level of review the source material requires.
- Source-linked research records
- Defined comparison groups
- Method notes and quality checks
- Research reports and visual summaries
This work supports research. It does not provide legal advice or predict the outcome of an individual case.
Also at ETHHSI Research
