> ## Documentation Index
> Fetch the complete documentation index at: https://docs.cognite.com/llms.txt
> Use this file to discover all available pages before exploring further.

# About diagram parsing for data modeling

> Extract information from static engineering diagrams to create interactive diagrams that you can explore in Cognite Data Fusion (CDF).

The process detects, extracts, and maps asset tags, file references, symbols, and connections from engineering diagrams to create a structured knowledge graph.

The parsing flow differs based on file characteristics:

**Rasterized files** support tag detection and mapping through text recognition and are limited to the Tag detection workflow.

**Vectorized files** support tag detection and mapping, symbol detection and library matching, connection tracing between symbols, and full detection workflow across all tabs.

<Note>
  *[Public preview](/cdf/product_feature_status#public-preview)*: In CDF, the **Full diagram parsing** option provides access to advanced parsing features across four tabs: **Tag detection**, **Symbols**, **Merge**, and **Connections**.

  Only the first page is parsed. If your multi-page PDF contains critical information on subsequent pages, consider splitting the file before parsing.
</Note>

This diagram illustrates the parsing flow for **vectorized** files:

```mermaid theme={"languages":{"custom":["/_languages/kuiper.json","../_languages/kuiper.json"]}}
flowchart TD
    A[<b>Preparations</b>: Setup libraries and files] --> B[<b>Parsing execution</b>: Text, symbols and connections]
    B --> C[<b>Verification and refinement</b>]
    C --> C1[<b>Tag detection tab</b>: Map assets and files]
    C --> C2[<b>Symbols tab</b>: Verify vectors]
    C --> C3[<b>Merge tab</b>: Link tags and symbols]
    C --> C4[<b>Connections tab</b>: Validate relationships]
    C1 & C2 & C3 & C4 --> D{Accuracy OK?}
    D -->|No| E[<b>Iteration and learning</b>: Refine and expand]
    E --> B
    D -->|Yes| F[<b>Output and integration</b>: Knowledge graph]
```

## Core concepts

Understanding these foundational concepts helps you work effectively with diagram parsing:

| Concept        | Definition                                                                                                                                                                  |
| -------------- | --------------------------------------------------------------------------------------------------------------------------------------------------------------------------- |
| **Connection** | A visual relationship between symbols, represented by lines or pipes in the diagram. Connections map how assets interact in your operations.                                |
| **Diagram**    | A single page of a parsed file containing all detected symbols and their connections. Each diagram is created using a specific library.                                     |
| **Geometry**   | One of the many possible visual compositions of a symbol. Each geometry is defined by SVG paths that help the algorithm recognize different visual styles.                  |
| **Instance**   | A detected occurrence of a symbol in the diagram. A single symbol can identify multiple instances based on the number of matching vectors found during the parsing process. |
| **Library**    | A CDF project-specific collection of symbols used to detect objects in engineering diagrams. You must select a library before running a parsing job.                        |
| **Merge**      | The process of linking detected tags and symbols that represent the same physical entity in the knowledge graph.                                                            |
| **Symbol**     | A blueprint that defines how to detect a particular type of equipment, such as a valve, instrument, or pipe. Each symbol contains one or more geometries.                   |
| **Tag**        | A text annotation detected in the diagram that identifies an asset or file reference. Tags can be linked to existing resources in CDF.                                      |
| **Template**   | A read-only set of symbols available by default in every CDF project. Templates provide a starting point for diagram parsing without requiring custom symbol creation.      |

<Note>
  For **Full diagram parsing** (*[public preview](/cdf/product_feature_status#public-preview)*), the following list clarifies terminology and deletion behavior.

  * A **symbol** in the library is the reusable definition. Each **detection** on a diagram is a matched **instance** you verify on the **Symbols** tab.
  * When **Merge** refers to **entities**, it means those diagram items as a single unit.
  * Deleting a library or symbol removes related detections across every diagram file the library is used on. Confirmation dialogs summarize the impact before you confirm deletion.
  * For deletion workflows, see [Manage symbol libraries](/cdf/integration/guides/contextualization/manage_libraries).
</Note>