Configure parsing scope and output format for Parse tasks.
The parsing configuration determines what content is processed, what auxiliary information is retained, and how audio/video content is organized. Select a file first, then adjust the settings relevant to your current goal.
Open ParseX, select "Document Parsing" on the left, and open the configuration area after selecting a file.
Coordinates answer "where the content is"; image descriptions answer "what the image expresses".
The term "story" on the page also applies to courses, training, and product demos, and refers to organizing media content. "Story Segments" and "Story Summary" require "Story Parsing" to be enabled at the same time.
Frame sampling selects frames from a video for understanding. The default is "Auto", where the system handles frame sampling; after selecting "Manual", you can specify the sampling method.
After adjusting, check the results using video segments that contain key operations.
"Output Parsing Results" determines the result files generated and supports multiple selections (for example, select Markdown file to generate a Markdown result file); "Table Output Format" determines how tables are expressed in the parsing results, with HTML or Markdown available.
The Markdown and JSON options in the result area are viewing modes; the settings here control the output content. You do not need to select JSON as the output file to view JSON. The API's file output options are described in a separate API reference.
Fill in supplementary instructions in "Custom Prompt" under Advanced Settings (optional, up to 4,000 characters).
After completing the configuration and submitting a parsing task, see Get Document Parsing Results.
Entry and Usage
Open ParseX, select "Document Parsing" on the left, and open the configuration area after selecting a file.
- Run once with the default settings on a familiar material.
- Compare the results and choose the items to adjust.
- Re-run after modifying, and check the effect of the adjustment.
- Once the effect meets your requirements, save the configuration for use with similar materials.
Text-and-Image Processing Settings
| Setting | Default | Purpose | When to adjust |
|---|---|---|---|
| Pages to parse | Empty; all pages are processed | Processes from page 1 up to the specified page | Fill this in when validating the beginning of a material first; for example, 3 means processing pages 1-3 |
| Parse headers and footers | Off | Reads content in headers and footers | Turn on when key information such as document numbers and versions is located in headers or footers |
| Return coordinates | On | Retains the positions of content in the original file | Keep on when the application needs to compare against the original by position |
| Image descriptions | On | Organizes what images express into text descriptions | Keep on when materials contain flowcharts, diagrams, or photos and the Agent needs to understand the images |
Audio and Video Processing Settings
| Page setting | Default | Purpose | Applicable needs |
|---|---|---|---|
| Speaker separation | On | Distinguishes different speakers in speech | Materials with multiple speakers, such as interviews and meetings |
| Story parsing | On | Organizes the overall content of the media | Understand what a recording or video is mainly about |
| Story segments | On | Organizes segments by content | Divide longer training or instructional materials into browsing units |
| Story summary | On | Extracts key points of the content | Let the application first determine the material's topic, then locate specific segments |
Video Frame Sampling
Frame sampling selects frames from a video for understanding. The default is "Auto", where the system handles frame sampling; after selecting "Manual", you can specify the sampling method.
| Manual setting | Meaning | What to watch when adjusting |
|---|---|---|
| Frame sampling rate | Number of frames selected per second | Whether demonstration steps or visual changes are preserved |
| Output frame image width | Horizontal size of frame images | Whether the output frames meet display or reading requirements |
| Output frame image height | Vertical size of frame images | Whether the output frames meet display or reading requirements |