Attachment Rule

This property defines the type of document that the rule will consider as a potential attachment. During evaluation, the rule is only applied to documents whose content type matches this value.

If left null, the rule is treated as a fallback and may apply to any document, including unclassified ones. This is useful for handling cases where attachments have not been typed or do not fit into a known schema.

This property defines the valid host document types to which an attachment may be affixed. The current document will only be attached to neighboring documents whose content type appears in this list.

If the list is empty, the rule permits attachment to any document type, regardless of classification. This allows for more general-purpose rules or fallback behaviors when host typing is inconsistent or unknown.

Can be one of the following values:

When AttachChildren is set to true, the attachment operation moves all children (folders and pages) of the attachment document to the host document, and then removes the original attachment folder. This is useful when the attachment is a container for related documents or pages that should remain grouped with the host after attachment. For example, if an email (the attachment) contains embedded files or sub-documents, enabling AttachChildren ensures that these items are also attached to the main document, preserving the logical structure and context.

If set to false, only the attachment folder itself is moved under the host document, leaving its children in place within the batch hierarchy. This allows for scenarios where the attachment should be grouped, but its contents should remain separate.

The behavior is implemented in the Attach activity: when AttachChildren is true, all children are moved and the attachment folder is removed; when false, only the folder is moved.

When set, the rule leverages this model to analyze document content and determine relationships based on natural language understanding.

This enables powerful semantic comparison beyond metadata and content type matching, especially for semi-structured or ambiguous attachments. ///

Can be one of the following types:

The Chat Parameters class defines a set of options for tuning the behavior of LLM completion requests in Grooper. These parameters allow users to control the randomness, diversity, and topicality of generated responses, enabling tailored outputs for a wide range of scenarios.

Typical usage involves adjusting one or more parameters to guide the model’s output toward the desired style or content. For example, increasing randomness for creative tasks, or reducing repetition for summarization.

Configuration Guidance

• Adjust either 'Temperature' or 'Top P' to control randomness and diversity, but not both simultaneously.
• Use 'Presence Penalty' and 'Frequency Penalty' together to reduce repetition and encourage broader topic coverage.

Usage in Grooper

Chat Parameters are used throughout Grooper to influence LLM completions for document processing, data extraction, and other AI-driven tasks.

Impact

Proper configuration of ChatParameters can significantly affect the quality, creativity, and reliability of LLM outputs. Experimentation is recommended to find the best settings for your specific use case.

Can be one of the following types:

If provided, this quoting method is used to extract relevant portions of a document's content to present to the LLM during evaluation. This helps reduce input size and improve accuracy by focusing the model on meaningful sections of the document.

If no quoting method is specified, the entire document content is sent to the model.

This is a plain-text prompt sent to the LLM, describing the logic it should use to decide whether the current document is related to its neighboring documents and should be treated as an attachment.

It should clearly explain the relationship criteria the model is expected to apply (e.g., explicit attachment/exhibit references, shared identifiers, semantic continuation, etc.).

This property is required when using AI-based attachment logic.