Technical Paper

Privacy and data risk

8.14 Surface

Yumo Yumo works with two data classes: user data and aggregate data. User data covers receipt content, spending history, and trust signals. Aggregate data covers price and basket-composition signals that enter the B2B data product.

SurfaceTechnical impactPublic control principle
User-data exposureIdentifiable receipt content is targetedOff-chain content and encrypted storage
Re-identificationAggregate data can match to an individual receipt or userk-anonymity and publication-group discipline
Legal data requestAn authority requests specific user dataPublished privacy policy and process record
Administrative accessThe operations team performs data-processing tasksTask-scoped access and audit trail

8.15 Control model

Off-chain receipt content. Receipt line items live in the off-chain ledger (04 §4.16). The on-chain layer carries bINT mint events and Merkle root commitments; content is processed in the data layer.

Aggregate publication discipline. The B2B data product follows k-anonymity and publication-group rules (05 §5.8). Publication groups are formed from region, category, and period cohorts with sufficient density.

Task-scoped access. Document-processing workers and administrative tools operate with the data scope needed for the relevant task. Retention, access, and deletion processes connect to the privacy policy and operational security process.

Audit trail. Administrative access records are retained for external review and internal-control cycles. Legal data requests are processed under the published privacy policy.

8.16 Evolution

Data-custody responsibility evolves with progressive localisation and regional-structure decisions. The architectural target stays stable: user receipt content remains off-chain, aggregate data becomes productised, and integrity proof is provided through on-chain commitment.