Powered by Zoomin Software. For more details please contactZoomin

Semaphore 5.13 Knowledge Model Management (KMM) How-To Articles

Enriching an Existing Model

  • Last Updated: November 20, 2026
  • 6 minute read
    • Semaphore
    • Version 5.13
    • Documentation


title: Enriching an Existing Model

Starting with Semaphore 5.13.0, you can use the Enrichment AI workflow to extend existing knowledge models with AI-generated concepts, relationships, labels, and metadata.

You can enrich an entire model, a concept scheme, an individual concept, or a concept and its descendants.

The Enrichment Agent uses a natural-language prompt and optional supporting documents to guide the generated content.

Before you begin

Before enriching an existing model:

  • Ensure that an administrator has configured the AI Model Builder and an LLM provider in Semaphore Studio. See Configure the AI Model Builder.
  • Ensure that you have permission to modify the knowledge model.
  • Identify the model, concept scheme, or concept you want to enrich.
  • Prepare a prompt describing the changes you want to make.
  • Optionally, prepare supporting documents containing relevant terminology, definitions, or domain-specific information.

Note: Enrichment automatically applies generated changes to the model or task in which it runs. The Enrichment Agent does not ask you to approve individual changes during execution.

Use a task when you want to review generated changes before incorporating them into the master model.

Enrichment procedure

Follow these steps to enrich an existing knowledge model:

  1. Open the model or task
  2. Start Enrichment
  3. Choose what to generate
  4. Enter your prompt
  5. Add supporting documents (optional)
  6. Run Enrichment
  7. Review the results

Step 1: Open the model or task

  1. Open Semaphore Knowledge Model Management (KMM).
  2. Open the knowledge model you want to enrich.
  3. If you want to review generated changes before committing them to the master model, create or open a task.

The selected model or task determines where the generated changes are applied.

Step 2: Start Enrichment

The way you start Enrichment determines its scope.

What you want to enrich How to start
Entire model Select Govern > Enrich Model.
Concept scheme Open the concept scheme's context menu and select Enrich.
Individual concept Open the concept's context menu, select Enrich, and choose Only the selected concept.
Concept and its descendants Open the concept's context menu, select Enrich, and choose Selected concept and its descendants.

For an entire model or concept scheme, no additional scope selection is required.

For a concept, choose whether to enrich only that concept or include its descendants.

Step 3: Choose what to generate

In the Enrichment Agent dialog, select the types of content you want to generate.

Option What it generates
Concepts New concepts within the selected scope.
Relationships Relationships between concepts.
Labels Labels for concepts.
Metadata Metadata values, such as definitions.

You can select multiple options in one enrichment operation, where available.

The available options depend on the selected scope.

For example, Concepts is not available when you select Only the selected concept. To generate additional concepts within an existing hierarchy, choose a scope that includes the concept's descendants.

Select only the options relevant to the changes you want to make.

Step 4: Enter your prompt

Enter a natural-language prompt describing the content you want to generate.

A useful prompt should specify:

  • The subject area or concepts to focus on.
  • The content to add or improve.
  • The terminology to use.
  • Any required definitions or relationships.
  • Constraints, such as preserving existing concepts or avoiding duplicates.

For example, to expand a renewable energy taxonomy:

Add three top-level concepts to the selected concept scheme:

    
  • Solar Energy
  • Wind Energy
  • Hydropower

Generate a clear, concise definition for each concept.

Use consistent terminology and avoid creating duplicates.

For this example, select Concepts and Metadata in the Enrichment Agent dialog.

The generated results depend on the prompt, existing model content, and configured LLM provider.

Step 5: Add supporting documents (optional)

You can upload supporting documents to provide additional context for the Enrichment Agent.

Supported file formats include:

Category File formats
Text and documents .txt, .md, .pdf, .docx
Semantic web and RDF .ttl, .nt, .n3, .trig, .nq, .rdf, .jsonld

Upload limits:

  • Up to five files.
  • Maximum file size of 4 MB per file.

Use documents that contain relevant terminology, definitions, requirements, or other domain-specific information.

Supporting documents can help guide generation, but they do not guarantee that every detail will be represented accurately.

Step 6: Run Enrichment

After selecting the enrichment options and entering your prompt:

  1. Review the selected scope.
  2. Confirm that the selected content types match your intended changes.
  3. Review the prompt and any supporting documents.
  4. Select Run enrichment.
  5. Wait for the Enrichment Agent to complete the operation.

Enrichment applies generated changes automatically. It does not request approval for each individual change.

Step 7: Review the results

When Enrichment finishes, inspect the affected model, concept scheme, or concepts.

Check the following:

Content What to review
Concepts Are new concepts relevant, correctly named, and free of unnecessary duplicates?
Relationships Are hierarchical and associative relationships accurate?
Labels Are preferred and alternative labels appropriate and consistent?
Metadata Are definitions and other generated values accurate and meaningful?
Existing content Have the intended meanings and relationships been preserved?

If you ran Enrichment within a task, review and correct the changes before incorporating them into the master model.

Important: Concepts with similar labels do not necessarily represent the same meaning. Consider their context, hierarchy, and intended use before deciding whether they should be reused, mapped, or represented separately.

Example 1: Add concepts to a concept scheme

Suppose you have a concept scheme for renewable energy and want to add concepts representing different energy sources.

  1. Open the knowledge model containing the concept scheme.

  2. Create or open a task if you want to review the changes before committing them.

  3. Open the concept scheme's context menu and select Enrich.

  4. Select Concepts and Metadata.

  5. Enter the following prompt:

    Create three top-level concepts:
    
          
    • Solar Energy
    • Wind Energy
    • Hydropower

    Add a concise definition for each concept.

    Use consistent terminology and avoid duplicates.

  6. Select Run enrichment.

  7. When the operation finishes, open the concept scheme.

  8. Verify that the expected concepts were created.

  9. Select each concept and review its definition in the Metadata section.

  10. Correct any inaccurate or unnecessary content.

  11. If you used a task, commit the reviewed changes to the master model when they are ready.

Example 2: Enrich an individual concept

Use this scope when you want to improve information associated with an existing concept without generating new concepts.

For example, you might want to add an alternative label or improve the definition of a concept.

  1. Open the model containing the concept.

  2. Create or open a task if appropriate.

  3. Open the concept's context menu and select Enrich.

  4. Choose Only the selected concept.

  5. Select Labels and Metadata.

  6. Enter a prompt such as:

    Review the concept "Solar energy".
    

    Add "Solar power" as an alternative label if it is not already present.

    Provide a clear, concise definition describing energy obtained from sunlight and its conversion into electricity or heat.

    Preserve the existing preferred label.

  7. Select Run enrichment.

  8. Open the concept and review its labels and metadata values.

Note: The Concepts option is not available when enriching only the selected concept.

Example 3: Enrich a concept subtree

Use this scope when you want to expand an existing hierarchy with additional concepts and relationships.

  1. Open the model containing the concept hierarchy.

  2. Create or open a task if appropriate.

  3. Open the parent concept's context menu and select Enrich.

  4. Choose Selected concept and its descendants.

  5. Select Concepts and Relationships.

  6. Enter a prompt such as:

    Review the renewable energy concept hierarchy.
    

    Expand its existing branches with relevant technologies and subcategories.

    Preserve the existing hierarchy and terminology.

    Avoid duplicate concepts and unnecessary relationships.

  7. Select Run enrichment.

  8. Review the generated concepts and relationships within the selected subtree.

Troubleshooting

The generated content does not meet your requirements

The quality of generated content depends on the configured LLM provider, the prompt, and the supporting information.

If the results are incomplete or inaccurate:

  1. Review the generated content.
  2. Identify missing, incorrect, or duplicate information.
  3. Correct the model content as necessary.
  4. Refine your prompt if you decide to run another enrichment operation.

A more specific prompt can help guide generation, but it does not guarantee accurate results.

Enrichment reports an error

An error does not necessarily mean that no changes were applied.

Some generated content may have been saved before the error occurred.

Before running Enrichment again:

  1. Refresh the affected model or concept view.
  2. Inspect the selected model, scheme, or concepts.
  3. Review the model history for any changes that may have been applied.
  4. Correct unwanted or inconsistent content.
  5. Retry only after confirming which changes are still needed.

This helps prevent duplicate concepts, labels, or metadata values.

The LLM provider rejects a request

Some LLM providers impose request or rate limits that can interrupt Enrichment.

If you encounter a provider-related error, contact your Semaphore administrator.

If the displayed error message does not provide enough information to identify the cause, the administrator can review the relevant Semaphore AI service logs.

Alert