Question reformulation with history in LangChain

Introduction

Question reformulation with history helps make follow-up questions clearer by using past conversation context. It improves understanding in chat or assistant apps.

When building a chatbot that remembers previous questions and answers.

When you want to turn short follow-up questions into full, clear questions.

When handling multi-turn conversations where context matters.

When improving user experience by avoiding repeated clarifications.

Syntax

LangChain

from langchain.chains import ConversationChain
from langchain.llms import OpenAI

conversation = ConversationChain(llm=OpenAI())

response = conversation.predict(input="Your question here")

Use ConversationChain to keep track of conversation history automatically.

The 'predict' method reformulates or answers questions using past context.

Examples

Asks a clear question without prior context.

LangChain

response = conversation.predict(input="Who is the president of the USA?")

Follow-up question that uses history to understand 'he' refers to the president.

LangChain

response = conversation.predict(input="Where was he born?")

Sample Program

This example shows how LangChain keeps track of conversation history. The second question 'Where was he born?' is understood using the answer to the first question.

LangChain

from langchain.chains import ConversationChain
from langchain.llms import OpenAI

# Initialize conversation with OpenAI model
conversation = ConversationChain(llm=OpenAI())

# First question
q1 = "Who is the president of the USA?"
answer1 = conversation.predict(input=q1)

# Follow-up question that depends on previous answer
q2 = "Where was he born?"
answer2 = conversation.predict(input=q2)

print(f"Q1: {q1}\nA1: {answer1}\n")
print(f"Q2: {q2}\nA2: {answer2}")

Important Notes

Make sure your OpenAI API key is set in your environment to use the OpenAI LLM.

ConversationChain automatically manages history, so you don't need to handle it manually.

Reformulated questions become clearer and more complete using past conversation context.

Summary

Question reformulation with history helps chatbots understand follow-up questions better.

LangChain's ConversationChain manages conversation history automatically.

This improves user experience by making conversations more natural and clear.