AI Extract Table Options

Defines options controlling how AI Extract works for a Data Table.

Remarks

The 'AI Extract Table Options' class provides advanced configuration for how Grooper's AI Extract method interacts with a specific Data Table. These options allow you to control the alignment strategy used by the LLM (Large Language Model) when extracting table rows, headers, and footers, ensuring that each row and cell is accurately located and mapped back to the document for review, navigation, and validation.

Role in Grooper

• Used to customize the extraction and alignment behavior for individual Data Tables when using AI Extract.
• Supports detection of table headers and footers, and enables row-level alignment for robust extraction from a wide range of tabular layouts.
• Integrates table-specific alignment instructions into the LLM prompt, improving extraction accuracy and reliability.

How It Works

• Each Data Table can be extended with an 'AI Extract Table Options' object, allowing you to override default alignment behavior.
• The 'Row Alignment' property determines what information is requested from the LLM (such as geometric regions or cell values) and how Grooper interprets the LLM's output to identify and align table rows in the document.
• Header and footer detection options control whether the LLM is asked to identify these special rows, and how they are mapped back to the document.

Configuration and Usage

• Choose a row alignment mode that matches the structure of your tables and the requirements of your extraction scenario.
• Enable header and footer detection as needed to improve accuracy for tables with variable or dynamic headers/footers.
• The selected options impact the JSON schema sent to the LLM, the expected response format, and the mechanism used to locate table rows and cells in the document.

Example Scenario

For an invoice line item table with dynamic headers, you might enable 'DetectHeader' and use 'Geometric' row alignment to have the LLM return bounding boxes for each row, allowing Grooper to precisely map each row and cell to its location in the document.

These options provide granular control over AI-driven table extraction, supporting robust, accurate, and context-aware data capture in Grooper.

Properties

Name	Type		Description
Row Alignment	RowAlignMode	►	Specifies the mode to be used for aligning captured table rows back to locations in the document. DefaultAuto Can be one of the following values: Name Value Description None 0 The table will not be aligned to the document. Only the table data is requested from the LLM; no alignment properties are included in the JSON schema. Table rows are used as-is, with no attempt to match or highlight them in the document. No navigation or highlighting will be available in the Data Grid or Data Viewer. Use this mode for simple tables where provenance is not important, or when alignment is not possible. Fastest extraction, but no review context or row-level navigation. ► Auto 1 Align to the document using the positions of extracted cell values. Grooper infers the position and boundaries of each table row by analyzing the positions of extracted cell values in the LLM response. No additional properties are requested from the LLM; alignment is inferred post-extraction. Supports navigation and highlighting of rows in the Data Grid and Data Viewer. Use this mode for tables where row boundaries can be reliably determined from cell positions, and explicit geometric regions are not needed. Enables advanced features such as multi-line row handling (see MultilineRows). ► Geometric 2 Ask the LLM to return a geometric region for each table row. This mode must be used in conjunction with the Layout Objects quoting method, which provides the location of each text segment. The JSON schema requests __pageNo and __bounds properties from the LLM for each row, specifying the page number and bounding box. Grooper uses this information to directly extract and highlight each row region from the document geometry. Most useful when the LLM or extraction process can provide precise location data, such as for visually distinct or structured tables. If the LLM returns invalid or missing coordinates, alignment will fail for that row. Use for advanced scenarios where spatial accuracy is critical, or when reviewing tables in visually complex documents. ► The 'Row Alignment' property determines how Grooper requests and interprets alignment information for table rows from the LLM. This setting controls what properties are included in the JSON schema, what the LLM is asked to return, and how table rows are mapped back to the document for navigation and review. If no human review is required, alignment may not be necessary. The selected mode impacts both the LLM prompt and the post-processing logic used to associate extracted rows with document locations. Choose the mode that best matches the requirements of your extraction and/or review scenario.
Detect Header	Boolean	►	If true, specifies that the LLM should detect the table header. DefaultFalse When enabled, the JSON schema requests header information from the LLM, such as header labels or geometric regions (depending on RowAlignment). Grooper uses this information to locate and map the table header in the document, improving accuracy for tables with variable or dynamic headers. Header detection is especially useful when column order or header text may change between documents. Only relevant when RowAlignment is not 'None'.
Detect Footer	Boolean	►	If true, specifies that the LLM should detect the table footer row. DefaultFalse When enabled, the JSON schema requests footer information from the LLM, such as footer labels or geometric regions (depending on RowAlignment). Grooper uses this information to locate and map the table footer in the document, supporting extraction and validation of totals or summary rows. Footer detection is useful for tables with calculated or validated totals. Only relevant when RowAlignment is not 'None'.
Multiline Rows	Multiline Row Settings	►	Controls how multi-line rows are detected for the purposes of table row alignment. EditorObject Properties Editor ConverterOn Off Converter Can be one of the following types: Value Description Enabled Disabled The 'Multiline Rows' property allows you to configure how the LLM and Grooper handle table rows that span multiple lines in the document. Only visible and relevant when RowAlignment is set to 'Auto'. Use this setting to improve extraction accuracy for tables where rows may wrap or contain embedded line breaks.

See Also

Multiline Row Settings