Data Master and Reference

Operator Horizonte registers customers in sales, contracts and billing. Two similar searches can be held by one person, by two families or by different companies. The master data government begins by deciding what an entity represents and who can change that decision.

Key concepts

The master data describe shared entities: client, product or supplier. The reference data delimits agreed values, such as contract states or document types. The transactions have events: an invoice, a payment or a contractual amendment. Linkage an invoice with a client does not turn the invoice into master data.

The identity should be stable and have a clear scope. The identifier using Sales may differ from that of Filling. A matching table keeps that relationship. The authorized source can also be replaced by an attribute: a commercial address does not necessarily have the same authority as a verified identification.

Client C-17 and P-2 plan are shared master identities. ACTIVO is a reference code with meaning as "valid contract." The F42 invoice is a transaction using those keys and code. To have an association with a teacher does not turn an invoice into a master: each object retains its identity, meaning and function within the exchange. ↗

Identity, Correspondency and Survival

Match identifies candidates who could represent the same entity. Merge It combines accepted records. The rules of survival determine what values and relationships survive. A high similarity score does not prove an identity: it should be interpreted with the available identifiers and the cost of an wrong union.

Horizonte defines matching rules, thresholds and cases that require review. The owner of Clients approves the policy. steward looks at ambiguities. applications retain the original identifiers and traceability. A consolidated registry needs to explain from what source each attribute proceeds and what rule it chose.

The name and phone share propose a competition between V-18 and F-07. The evidence decides if they represent the same identity. When confirmed, values are chosen by attribute with their rule and source. If the evidence is negative or insufficient, it does not fuse automatically: it is rejected with evidence or the review remains pending. ↗

For reference data, the risk may be different: a code valid for a system does not have to be valid for another period or market. The lists require owner, definition, effective date and mapping. Retiring "suspended" to "inactive" without agreeing its meaning can alter care's indicators and decisions.

Data Shared with Oracle

Oracle AI Data Platform and Oracle Autonomous AI Lakehouse allow them to prepare, link and consume shared datasets. That prosecution requires a clear identity policy. saving a client board into a Gold cape doesn't automatically turn it into a reliable master.

In Oracle Fusion Cloud Enterprise Data Management, sets up the rules of correspondence of the domain, checks the candidates and applying survival rules to choose the values of the consolidated registry. For Horizonte, steward contrasts Sales and Filling Identifiers. The approved rule lays down what source each attribute brings. The result preserves the chosen values and matching with the registration of origin so that consumers can link their transactions with their agreed identity.

The distribution of the approved set requires export and set loads. For example, Oracle Fusion Cloud Enterprise Data Management allows export Universal Dimension Correlation to CSV. The integration preserves approved identifiers, versions and rules. The diagram does not require automatic synchronization with consumer platforms.

Oracle Fusion Cloud Enterprise Data Management is shown as a SaaS area with its own scope and requirements. Correspondency, review and survival values produce an approved set with identity, code and version. Configure exports and loads can get that set to Oracle AI Data Platform and Oracle Autonomous AI Lakehouse. Check the identity decision and each integration before posting for consumption. ↗

The process evidence includes revised candidates, fusion decision, survival values, identifier mapping and published version. A useful quality metric is the rate of mergers reversed by error and with revised volume. maximizing mergers by hour can give an incentive to wrong decisions.

Explanated example

Sales have "Ana Pérez," V-18 identifier, and Filling to "Ana M. Pérez, "F-07 identifier. They share a phone but that number belongs to a family. The phone number identifies a candidate match; it is not sufficient to merge them.

The steward checks a verified identification and confirms that both records represent the same client. The approved rule retains the valid source identification, the most recent confirmed contact address and both source identifiers. A third person with the same phone remains separate.

Anne's identity's been confirmed. The table explain what value survives by attribute. Identification comes from a validated source, the most recent confirmation address and Sales and Filling identifier are preserved as matching. A phone share remains a sign. Consent requires review by purpose and validity, separate from the consolidation of identity. ↗

The conjunctions are reviewed independently by purpose and validity. A merger of identity does not permit an extension of billing consent to a commercial campaign. Horizonte published the teacher and his or her correspondent and consumers check that bills are still linked to the right entity.

Common

• "Same name means same client." Homonyms and incomplete data require additional evidence.
• "The last record always wins." The novelty does not prove authority. a recent modification may be incorrect.
• "Gold already means master." A processing layer does not replace rules of identity, approval and distribution.

Decision exercise

Two clients share name, address and phone, but their verified identifications are different. Marketing requests to join them to reduce duplication. He decides how to deal with his candidacy, what evidence to preserve and who resolves his disagreement. Consider what would happen with bills and conjunctions.