[Agent Scenario Request]: Missing Graph Description When Using Paper Parsing Method

[nikhilmgeorge7]

Self Checks

• I have searched for existing issues search for existing issues, including closed ones.
• I confirm that I am using English to submit this report (Language Policy).
• Non-english title submitions will be closed directly ( 非英文标题的提交将会被直接关闭 ) (Language Policy).
• Please do not modify this template :) and fill in all the required fields.

Is your feature request related to a scenario?

Describe the feature you'd like

When using the Paper parsing method in RAGFlow, the graph description field is not generated.
However, when using the default/general parsing method, the graph description appears as expected.

Steps to Reproduce

Upload a document and choose Paper as the parsing method.

Process the document.

Open the graph section.

Observe that the graph description is missing.

Repeat the process using the General parsing method and notice that the graph description is generated normally.

Expected Behavior

The graph description should be generated consistently regardless of whether the Paper or General parsing method is used.

Actual Behavior

Paper parsing: No graph description is generated.

General parsing: Graph description is present.

**

**

It is unclear whether this is a limitation of the Paper parser or an unintended regression. Any clarification or fix would be appreciated.

Documentation, adoption, use case

Additional information

No response

[dosubot]

Hi @nikhilmgeorge7! I'm Dosu and I’m helping the ragflow team.

The missing graph description when using the Paper parsing method is not an intentional limitation or a recent regression. Instead, it's almost always due to configuration or compatibility issues with the vision (image-to-text) model required for figure/graph description generation.

The Paper parser relies on a vision model (IMAGE2TEXT) to generate graph descriptions. If this model isn't configured or available for your tenant, the enhancement step that produces graph descriptions is silently skipped see code. This can happen if the global or tenant-specific image-to-text model settings are missing, misconfigured, or if the selected model/provider is incompatible settings reference.

To resolve this, check that:

• The global and tenant-specific IMAGE2TEXT model is set up and authorized (see IMAGE2TEXT_MDL, IMAGE2TEXT_CFG, your tenant's img2txt_id, and TenantLLM entries).
• The selected vision model is compatible with your deployment (some providers or models may not support all required arguments; switching to a different vision model can resolve errors like unexpected keyword arguments example).
• If you encounter errors during figure extraction, review logs for model invocation errors and try a different vision model if needed.

There are no explicit flags to enable or disable graph description generation for the Paper parser—it's entirely governed by the availability and correct configuration of the vision model. If your configuration appears correct but the issue persists, keep an eye out for updates or fixes in future releases, as improvements to error handling and compatibility are ongoing.

To reply, just mention @dosu.

---

^{How did I do? Good | Irrelevant | Incorrect | Verbose | Hallucination | Report 🐛 | Other}