Croissant adds clear rules to check dataset use restrictions automatically
A Policy Profile for Croissant: Refusal as a Property of the Dataset
Machine LearningCryptography and SecurityDatabases
Summary
Datasets often have rules about how they can be used, but previously these rules weren’t always easy for software to check or understand clearly. The authors worked on Croissant, a way to write these dataset rules so that computers can automatically decide if an operation is allowed or refused. They created a clear system with exact steps and recorded decisions to make sure checks are reliable and fast. This helps systems that manage datasets follow restrictions correctly and transparently.
What this means in practice
- •For data platform engineers: Enable automated enforcement of dataset use rules to block unauthorized actions directly from dataset metadata.
- •For workflow automation developers: Integrate precise policy checks into pipelines that handle datasets to ensure compliance without manual intervention.
Authors
Alexander Chernov
Abstract
Croissant is the de facto machine-readable descriptor for ML datasets: JSON-LD over schema.org. Since version 1.1 it also carries data use conditions, recommending DUO and ODRL for them. What no version specifies is how any of them is evaluated: no decision procedure, no bound on evaluation cost, no outcome for a condition an implementation cannot evaluate, no record of what was checked, and nothing on composition with caller-side authority. We supply that half. An additive profile lets a dataset declare the operations it admits and the conditions under which it admits them, over a closed set of five operators whose decision procedure is given in full, so a gate decides from the descriptor alone and records what it checked. Two corpora evaluate it and their evidence is kept apart. Three descriptors that gated a real nf-core pipeline give the deployment result: decisions from a profile document match the gate's native descriptor record for record, stripping the layer leaves a valid Croissant document, and the added cost is 11.7 $μ$s against a 119 $μ$s decision. A corpus generated from the profile's grammar gives the breadth, covering every operator, refusal class and conformance clause. Across its valid cases, 552 complete decision records agree three ways -- native descriptor, profile terms, and the same policy as ODRL in usageInfo. The carrier is therefore not the contribution; the evaluation semantics is. Finally, caller-bound and data-bound policies range over non-overlapping state spaces, so neither permit set contains the other.