Question and Answer

Using the /retreival endpoint, you can build your own legal Question and Answer with sources chain by consuming the helpmefindlaw API as a retreiver with langchain or llama index. This is usefull if you want to get access to the raw retreival functionality of our RAG pipeline, which includes the joint functionality of a vector search and knowledge graph over our custom laws dataset.

Langchain Example

We define a custom langchain retriever like so

examples/src/utils/langchain.py
from typing import List

from .api import HelpMeFindLawRetrievalInput, HelpMeFindLawClient
from langchain.callbacks.manager import CallbackManagerForRetrieverRun
from langchain.schema import BaseRetriever, Document


class HelpMeFindLawRetriever(BaseRetriever):
    """`HelpMeFindLaw` retriever.

    It is used for the question-answering with sources with
    the `/retrieve` endpoint of HelpMeFindLaw."""

    client: HelpMeFindLawClient

    def _get_relevant_documents(
        self, query: str, *, run_manager: CallbackManagerForRetrieverRun
    ) -> List[Document]:
        """Get documents relevant for a query."""
        docs = self.client.retrieve(HelpMeFindLawRetrievalInput(prompt=query))

        # parse source nodes
        docs = []
        for doc in docs:
            metadata = doc.metadata or {}
            docs.append(
                Document(
                    page_content=doc.text,
                    metadata=metadata)
            )
        return docs

With this retriever, it is quite straight forward to add HelpMeFindLaw into your langchain workflows. The example below builds a RetrievalQA chain from the HelpMeFindLaw retriever. This allows you to develop your own legal QandA system leveraging the data and retreival optimizations within HelpMeFindLaw.

qanda.py
import os
from dotenv import load_dotenv
from utils.api import HelpMeFindLawClient
from utils.langchain import HelpMeFindLawRetriever
from langchain.chains import RetrievalQA
from langchain.chat_models import ChatOpenAI

client = HelpMeFindLawClient(token="******")
retriever = HelpMeFindLawRetriever(client=client)
llm = ChatOpenAI(model_name="gpt-3.5-turbo", temperature=0)
chain = RetrievalQA.from_chain_type(llm, retriever=retriever)

question = "What is the statute of limitations for a personal injury claim in California?"
result = chain({"query": question})

Question and Answer

Langchain Example​

Langchain Example