Data Management & Integration Flashcards

During customer matching, records 'Jon Smyth, 12 Oak St' and 'John Smith, 12 Oak Street' should be linked. Which technique is most appropriate?

Answer: Probabilistic (fuzzy) matching with weighted scores

Probabilistic matching scores similarity across attributes, so it can link near-duplicate records that exact matching would miss.

What is the primary purpose of a 'golden record' in master data management?

Answer: A single trusted, consolidated version of a master data entity

A golden record is built from multiple sources using survivorship rules to give the best version of the truth.

Survivorship rules in MDM determine what?

Answer: Which source attribute values populate the consolidated master record

Survivorship rules such as most recent, most trusted source, or most complete pick which value wins for each attribute.

An enterprise has dozens of point-to-point interfaces that are costly to maintain. Which integration architecture most directly reduces this complexity?

Answer: Hub-and-spoke or enterprise service bus / event broker

A hub or broker cuts n(n-1)/2 direct connections down to roughly one connection per system.

Reference data differs from master data mainly because reference data:

Answer: Consists of code sets and classifications used to categorize other data

Reference data, such as country codes, currency codes, and status values, defines allowed values for categorizing other data.

In event-driven integration using a platform like Apache Kafka, what does the publish/subscribe model provide?

Answer: Loose coupling where producers don't need to know consumers

Pub/sub decouples producers from consumers, so new subscribers can be added without changing the source.

Which artifact best documents how a data element flows from source systems through transformations to a report?

Answer: Data lineage map

Data lineage traces where data originates, how it is transformed, and where it is used.