Golden glossary

Definitions of the terms used in Golden configuration, data processing, and APIs.

Bucket

One index’s contribution to duplicate detection: the records sharing one key of one indexer mapping. A cluster is made of one or more buckets, and the API has a separate operation family for each. See also cluster.

Classification

The customer-visible outcome assigned to a candidate group, such as MATCH, NON_MATCH, REVIEW, or IGNORE.

Classifier

Resource that assigns a customer-visible classification to duplicate candidates from configured comparisons and thresholds.

Cluster

The logical group a person decides on, assembled from one or more buckets. It reports the bucketIds it was built from. See also bucket.

Data scope

The restriction that says which entities, and which rows of an entity, a principal may read. Independent of the role, which says what kind of operation they may perform; both have to permit an operation.

Data view

Reusable presentation configuration for record titles, lists, summaries, and forms. It is written against a dataset’s columns and chosen by the entity that owns the table.

Dataset

The reusable definition of a record’s fields, data types, identity, and validation behavior.

Destination

A supported output configured to receive records or product events from a data flow. Some interfaces use sink for this concept.

Entity

The Golden configuration for one business subject and its table, capabilities, resources, sources, and destinations.

Escalation

A recorded state on a duplicate case passed to somebody with more context. It carries who escalated it, when, and the note they left, and the case moves to its own queue. Not a handoff convention outside the product.

Expert mode

The duplicates screen that lists every group of a data set in any state, with filters and bulk actions. It is available to ADMIN, whose Duplicates link opens this mode. STEWARD and VIEWER use Simple mode, which presents one case at a time; VIEWER cannot submit decisions.

Golden record

The mastered representation created after related records are resolved.

Indexer

Resource that defines searchable keys and which records can become duplicate candidates.

Merger

Resource that defines precedence for values in a mastered record.

Pipeline

An ordered set of configured processing steps applied during a data flow.

Product event

A customer-visible delivery item associated with a configured destination. It is not an infrastructure-monitoring event.

Quality score

An integer from 0 to 100 summarizing how far one record satisfies its dataset’s rules. The score measures conformance to configured rules; it does not establish that a record represents the intended subject.

Record

One representation of a business subject, shaped by a dataset and stored in a table.

Resource

Reusable Golden configuration for a dataset, source, destination, transformation, pipeline, search, resolution, or stewardship behavior.

Sample project

An installable configuration with synthetic data for learning Golden. Aurelia Utilities is the sample used in these guides, available in Settings.

Scope column

The entity column whose values partition its records for row-level access. A data scope grants a principal named values of it.

Separation

The recorded judgment that two records are not the same subject. It survives reindexing, and a later merge of that pair is refused unless somebody crosses it explicitly and gives a reason.

Snapshot

A saved aggregate of a table’s record quality for a closed UTC day. Daily snapshots support the quality trend and remain distinct from on-demand measurements.

Table quality measurement

A saved observation of a table’s quality and issue counts at an actual instant, requested by an administrator or created during sample installation. Measuring aggregates stored record measurements; it does not recalculate them. The Quality screen can display this observation or a daily snapshot. Opening the screen does not create another measurement.

Source

A supported input configured to read records into a Golden data flow.

Steward

A person with responsibility for data decisions, or configured Golden behavior that applies approved stewardship policy. The context makes the meaning clear.

Table

A managed collection of records shaped by a dataset.

Task

One asynchronous execution with progress, status, messages, and a terminal outcome.

Transformation

Reusable configuration that maps or changes records between data structures.