recap.rag

Submodules

recap.rag.base

Base class for RECAP RAG to be used by the retrieval instance

class recap.rag.base.BaseRAG

Bases: object

recap.rag.mred

class recap.rag.mred.MREDReviewRAG(db_path: str = './chroma')

Bases: object

This class hanldes the data processing and RAG process for the MRED reviews dataset It handles ingestion and retrieval

ingest_data(directory: str | Path, batch_size: int = 128, clean: bool = False) → None

Ingest MRED dataset from the specified directory.

Parameters:

• directory – The folder directory containing full_data_info.txt and review files

• batch_size – Batch size of reviews to ingest in vector store. Defaults to 128

• clean – Deletes the collection in chroma. Defaults to False

retrieve(query: str | None = None, filter=None, k: int = 10) → tuple[str, list[dict]]

Retrieve information to help answer a query.

Parameters:

• query – The folder directory containing full_data_info.txt and review files Defaults to None

• filter – A valid chroma Where filter dictionary object. If None provided no metadata filtering will be done. Defaults to None

• k – This is number of entries to return from a similarity search. It is ignored if no query provided

validate_filter(filter) → str | None

Module contents