⚙️ Metadata Naming Guidance¶

Deterministic, project-specific naming assistance for TargetColumn modeling based on existing metadata mappings, not AI or global naming dictionaries.

🔧 1. Purpose¶

Metadata Naming Guidance helps modelers reuse already established column naming decisions while editing TargetColumns.

It is designed for practical modeling situations where technical source-system fields need to be translated into consistent platform names.
For example, SAP-style source columns such as MANDT, VBELN, KUNNR or ProductModelID may need to become project-specific names such as client_no, sales_document_no, customer_no or product_model_id.

The goal is not to define one universal naming standard.
The goal is to make already used project-specific naming decisions visible at the moment where a modeler needs them.

🔧 2. What it does¶

Naming Guidance analyzes existing metadata mappings and shows known target names for the current input column.

When a modeler edits a TargetColumn name, elevata can show:

Naming guidance  ProductModelID

Recommended: product_model_id

Known target names
product_model_id  used 1 time
  rawcore.rc_aw_product_model

If the recommendation is clear, the modeler can apply it directly from the inline editor.

The existing TargetColumn rename flow remains authoritative. Applying a recommendation still uses the standard rename validation, collision checks, audit handling and former-name tracking.

🔧 3. How guidance is derived¶

Naming Guidance is deterministic and read-only while building suggestions.

It uses existing TargetColumnInput metadata.

🧩 3.1 Direct source inputs¶

For direct source inputs, guidance uses:

SourceColumn.source_column_name

🧩 3.2 Upstream target inputs¶

For upstream target inputs, guidance uses:

upstream TargetColumn.target_column_name

The lookup intentionally follows only one column-lineage edge.

This keeps guidance aligned with the current modeling step:

Source system column: AccountNumber
rawcore column:       account_no
bizcore input:        account_no

When editing a bizcore column based on account_no, the relevant input name is account_no, not the original AccountNumber.

🔧 4. Guidance states¶

🧩 4.1 new¶

No previous target name is known for the current input column.

The UI keeps this state quiet to avoid unnecessary noise.

🧩 4.2 suggested¶

There is either one known target name or a clearly dominant target name.

The recommendation is shown in the TargetColumn inline editor and can be applied directly.

🧩 4.3 conflict¶

There are multiple known target names without a clear dominant naming pattern.

In this case, elevata shows the known variants and examples, but does not choose a recommendation.

🔧 5. Scope and boundaries¶

Naming Guidance focuses on technical naming decisions in modeling layers:

rawcore
bizcore

The following are intentionally excluded from recommendation evidence:

serving-layer friendly names
historized rawcore datasets
ambiguous mixed-input evidence where one previous target column cannot be clearly attributed to one input name
system-managed technical columns

For multi-input TargetColumns, guidance is shown separately per immediate input column when clear previous usage exists.

This keeps the guidance focused on reusable technical naming conventions rather than presentation-layer labels, generated structures or ambiguous derivation evidence.

🔧 6. What it does not do¶

Naming Guidance does not:

enforce names
block saves
mutate metadata while building guidance
introduce AI-based inference
use a global dictionary
require a glossary workflow
add database tables or migrations

It is a transparent advisory layer derived from the metadata that already exists.

🔧 7. Why this matters¶

Consistent naming is difficult when source systems use cryptic technical identifiers.

The challenge is not only choosing a good name once.
The challenge is remembering how the same field was named elsewhere.

Naming Guidance reduces that manual lookup effort without replacing modeler judgment.

It makes project-specific naming conventions visible where they are needed most: directly in the TargetColumn modeling workflow.