Extracted

Can be one of the following types:

The 'Quote Extractor' property allows you to specify a Value Extractor that targets specific content within a document to be included as the quote for the AI model.

• When set, all values matched by the extractor are concatenated and supplied as the document quote.
• If left unset, the entire document (or extraction scope) is quoted, making this method suitable for both targeted and full-document quoting.

Configuration Guidance

• Use a Value Extractor to focus the AI on key fields, patterns, or regions, such as invoice numbers, dates, or specific sections.
• Common extractor types include regular expressions, list matches, or references to shared extractors.
• For reusable logic, configure a Reference extractor that points to a shared Value Extractor object.

Impact of Changes

• Specifying a targeted extractor reduces prompt size and can improve AI accuracy by focusing on the most relevant content.
• Leaving this property unset results in the entire document being quoted, which may increase token usage and cost.

Example

To extract all invoice numbers from a document, configure a regular expression extractor such as:

Invoice\s*No\.?:?\s*(\d+)

All matches will be concatenated and provided to the AI as the quote.

The 'Result Delimiter' property determines the character or string used to separate multiple values extracted by the Value Extractor when assembling the quoted text for the AI model.

Purpose

• When the Value Extractor matches more than one value, each result is joined using the specified delimiter.
• This allows you to control how extracted values are presented—either as a list, a single line, or with custom separators.
• Common delimiters include newline (\n), tab (\t), comma (,), or any custom string.

Configuration Guidance

• Set 'Result Delimiter' to \n (the default) to place each extracted value on a new line.
• Use \t to separate values with tabs, which is useful for tabular data or when combining with tab marking.
• Specify a comma, semicolon, or other character for inline lists or custom formatting.
• Escape control characters using standard notation (e.g., \n for newline, \t for tab).

Impact of Changes

• Changing the delimiter affects how the quoted text is formatted and interpreted by the AI model.
• Newline or tab delimiters improve readability and support extraction of structured data.
• Inline delimiters (such as commas) may reduce prompt size but can make it harder to distinguish individual values.

Example

If Select Page is used to extract the first and last 5 pages of a 20-page document, and the 'Result Delimiter' is set to \f, extractor hits will be separated by form feed characters:

The 'Preprocessing' property enables advanced manipulation of the document's text before extraction or quoting, improving the AI model's ability to interpret complex layouts, tabular data, or natural language.

• Preprocessing can include paragraph marking, tab marking, vertical tab marking, and control character handling.
• These options help normalize the document's structure, making it easier for extractors and AI models to match values that span lines, columns, or irregular formatting.

Configuration Guidance

• Enable Paragraph Marking to merge wrapped lines within paragraphs, allowing extraction of values that span multiple lines.
• Use Tab Marking to insert tab characters at large horizontal gaps, distinguishing between normal spaces and significant field separators.
• Adjust Vertical Tab Marking for documents with multi-column or tabular layouts.
• Use the 'Ignore Control Characters' option to remove or replace problematic whitespace or line breaks.

Impact of Changes

• Proper preprocessing can significantly improve extraction accuracy and AI understanding, especially for documents with inconsistent formatting.
• Overly aggressive preprocessing may remove important structure; test settings to ensure optimal results.

Example

For a table with columns separated by large spaces, enable tab marking so that extractors can use \t in regular expressions to match column boundaries.

For natural language documents, enable paragraph marking to allow extraction of values that span multiple wrapped lines within a paragraph.

The 'Document Level' property determines the scope of extraction for quoting.

• When enabled (true), extraction is performed on the entire Document Instance.
• When disabled (false), extraction is limited to the current Section Instance or other Field Container Instance.

Configuration Guidance

• Enable this property when you want to extract or quote content from the entire document, outside the scope of the current Field Container Instance.
• Disable to restrict extraction to a specific section, reducing prompt size and focusing the AI on a narrower context.

Impact of Changes

• Enabling document-level extraction increases the amount of content provided to the AI, which may improve context but also increase token usage.
• Disabling may improve performance and reduce cost, but risks omitting important context if the extraction scope is too narrow.

The 'Include Line Numbers' property controls whether each line of the extracted or quoted text is prefixed with its line number before being supplied to the AI model.

Purpose

• Adds a numeric prefix to each line, making it easier to reference, review, or discuss specific lines in the quoted content.
• Useful for prompt engineering, debugging, or when the AI model needs to refer to line positions in the source text.

Configuration Guidance

• Enable (true) to add line numbers to each line of the quoted text.
• Disable (false) to output the quoted text without line numbers.

Impact of Changes

• Enabling this property increases the length of the quoted text, which may affect token usage and prompt size.
• Line numbers can improve clarity when reviewing AI responses or when instructing the model to reference specific lines.

Example

With 'Include Line Numbers' enabled, the quoted text will appear as:

01: Invoice Number: 12345
02: Date: 2024-06-01
03: Total: $1,234.56

With this property disabled, the same text would appear as:

Invoice Number: 12345
Date: 2024-06-01
Total: $1,234.56

The 'Paginated' property controls whether the quoted content is divided into individual pages when supplied to the AI model.

Purpose

• When enabled (true), the extracted or preprocessed text is split into pages, and each page is quoted separately.
• Each page is prefixed with its page number, making it easier to reference, review, or discuss specific pages in the quoted content.
• This is especially useful for documents with multiple pages, such as contracts, reports, or forms, where page context is important.

Configuration Guidance

• Enable 'Paginated' to ensure the AI model receives page-separated content, preserving the original document structure.
• Disable to supply the entire quoted text as a single block, regardless of page breaks.
• Use in combination with 'Include Line Numbers' to further enhance clarity and navigation within each page.

Impact of Changes

• Enabling this property increases the clarity and navigability of quoted content, especially for multi-page documents.
• Page separation may increase prompt size and token usage, but improves the ability to reference specific pages in AI interactions.
• Disabling may reduce prompt size but can make it harder to locate or discuss content from specific pages.

Example

With 'Paginated' enabled, the quoted text will appear as:

Page 1

Invoice Number: 12345
Date: 2024-06-01
Total: $1,234.56

Page 2

Notes: Payment due in 30 days.

With this property disabled, the same text would appear as:

Invoice Number: 12345
Date: 2024-06-01
Total: $1,234.56
Notes: Payment due in 30 days.

Overview

The 'Name' property provides a concise, human-readable label for the quoting method. This value is used in Grooper’s user interface to help users identify, select, and configure quoting strategies. It is also injected into the prompt context sent to the large language model (LLM), where it serves as a section header or explicit label for the quoted content.

LLM Context Injection

• When the output of multiple quoting methods is combined (as in Multi Quote), the 'Name' is used to clearly distinguish each section of quoted content for the LLM. This helps the model understand that different representations or perspectives of the same document are being provided.
• In quoting methods such as Layout Objects, the 'Name' may be included in the prompt to clarify the structure or intent of the data, guiding the LLM’s interpretation and extraction logic.
• A descriptive and unambiguous name improves the LLM’s ability to reference, compare, and utilize the quoted data accurately, especially in complex or multi-view prompts.

Configuration and Usage

• Choose a name that reflects the quoting method’s purpose, formatting, or the type of content it produces.
• If left blank, the quoting method’s type name will be used as a default.
• Use consistent naming conventions to support clarity for both users and the LLM.

Example

• Claim Data
• Table Headers
• Layout Objects (JSON)

Best Practices

• Use names that are meaningful to both users and the LLM.
• When using Multi Quote, ensure each quoting method has a unique and descriptive name to maximize clarity in the combined prompt.
• Avoid generic names; be specific about the quoting method’s approach or output.

Overview

The 'Description' property provides a detailed, plain-language explanation of the content or data injected by this quoting method. It is intended to help the LLM understand the structure, semantics, and intended use of the quoted content, as well as to guide the model’s behavior when processing or extracting information from it.

LLM Context Injection

• The description is included directly in the prompt context sent to the LLM, typically as a prefix or annotation before the quoted data.
• In Multi Quote, each quoting method’s description is presented alongside its output, giving the LLM explicit instructions or context for interpreting each section.
• In Layout Objects, the description can clarify the meaning of structured layout data, JSON schemas, or other specialized formats, helping the LLM understand how to use the data for extraction or analysis.
• A well-crafted description can specify the format, highlight important features, explain relationships, or instruct the LLM on how to process the data (e.g., “This section contains a JSON array of line items, each with fields for description, quantity, and price. Extract each item as a separate record.”).

Configuration and Usage

• Write the description for the LLM, focusing on what the data is, how it is structured, and what the model should do with it.
• Include details about formatting, delimiters, field names, or any special conventions used in the quoted content.
• Use markdown formatting, examples, or sample data to illustrate the structure or intent.
• Update the description if the quoting method’s configuration or output changes.

Example

This section contains a table of invoice line items, formatted as plain text with tab-separated columns.
Each line represents a single item, with columns for description, quantity, and unit price.
Extract each line as a separate record and map the columns to their corresponding fields.

Best Practices

• Be explicit and unambiguous; avoid assumptions about the LLM’s prior knowledge.
• Tailor the description to the specific quoting method and output format.
• When using Multi Quote or Layout Objects, ensure each description is distinct and clearly describes the associated data.
• Use examples to illustrate complex or non-standard formats.