components.result

source

This module provides the LettuceResult class, which is designed to hold and manage data flowing through different stages of the lettuce pipeline.

LettuceResult

class LettuceResult(
	search_term: str
)

A class to hold the results of different stages of lettuce pipelines.

Parameters

Methods

• search_term: str

  The search or source term serving as input for the pipeline (e.g., informal drug name)

add_vector_search_results

def add_vector_search_results(
	vector_search_results: 
)

Stores vector search results from embedding-based similarity search part of the RAG pipeline.

Parameters

• vector_search_results: List[Dict[str, Any]]

  List of dictionaries containing vector search results, typically with similarity scores and matched concepts

add_llm_answer

def add_llm_answer(
	llm_answer: str
)

Insert the results of an LLM assistant’s inference.

Parameters

• llm_answer: str

  The reply of an LLM

get_query

def get_query()

Retrieve the appropriate part of the result object for querying an OMOP-CDM database. If no previous stages have been executed, uses the search_term. If there’s only a vector search result, uses the top result. If there is a response from an LLM, uses that.

Returns

• str

  Term for a database query

add_matches

def add_matches(
	omop_matches: list,
	threshold: float
)

Inserts the matches retrieved from a database search, after fuzzy string filtering.

Parameters

• omop_matches: list

  A list of the matches retrieved from the database

• threshold: float

  The threshold used for filtering

to_dict

def to_dict()

Serialises the result as a dictionary

Returns

• dict

  Pipeline results serialised