compass.validation.content.LegalTextValidator#

class LegalTextValidator(tech, doc, *args, score_threshold=None, **kwargs)[source]#

Bases: TextKindValidator, JSONFromTextLLMCaller

Parse chunks to determine if they contain legal text

Parameters:

  • tech (str) – Technology of interest (e.g. “solar”, “wind”, etc). This is used to set up some document validation decision trees.

  • doc (elm.web.document.BaseDocument) – The document being validated. This is used to set up some document validation decision trees and should contain metadata about the document (e.g. whether or not it was extracted from OCR).

  • score_threshold (float, optional) – Minimum fraction of text chunks that have to pass the legal check for the whole document to be considered legal text. If None, uses a custom threshold that caps at 0.8 for documents with a lot of content. By default, None.

  • *args – Parameters to pass to the JSONFromTextLLMCaller initializer.

  • **kwargs – Parameters to pass to the JSONFromTextLLMCaller initializer.

Methods

call(sys_msg, content[, usage_sub_label])

Call LLM for structured data retrieval

check_chunk(chunk_parser, ind)

Check a chunk at a given ind to see if it contains legal text

Attributes

SYSTEM_MESSAGE

System message for legal text validation LLM calls

is_correct_kind_of_text

True if text was found to be from a legal source

score_threshold

Threshold for validation check
SYSTEM_MESSAGE = 'You are an AI designed to classify text excerpts based on their source type. The goal is to identify text that is extracted from **legally binding regulations (such as zoning ordinances or enforceable bans)** and filter out text that was extracted from anything other than a legal statute for an existing jurisdiction.'#

System message for legal text validation LLM calls

property is_correct_kind_of_text#

True if text was found to be from a legal source

Type:

bool

property score_threshold#

Threshold for validation check

Type:

float

async check_chunk(chunk_parser, ind)[source]#

Check a chunk at a given ind to see if it contains legal text

Parameters:

  • chunk_parser (ParseChunksWithMemory) – Instance that contains a parse_from_ind method.

  • ind (int) – Index of the chunk to check.

Returns:

bool – Boolean flag indicating whether or not the text in the chunk resembles legal text.

async call(sys_msg, content, usage_sub_label=LLMUsageCategory.DEFAULT)#

Call LLM for structured data retrieval

Parameters:

  • sys_msg (str) – The LLM system message. If this text does not contain the instruction text “Return your answer as a dictionary in JSON format”, it will be added.

  • content (str) – LLM call content (typically some text to extract info from).

  • usage_sub_label (str, optional) – Label to store token usage under. By default, "default".

Returns:

dict – Dictionary containing the LLM-extracted features. Dictionary may be empty if there was an error during the LLM call.