TF-IDF Retriever

This notebook goes over how to use a retriever that under the hood uses TF-IDF using scikit-learn.

For more information on the details of TF-IDF see this blog post.

from langchain.retrievers import TFIDFRetriever

# !pip install scikit-learn

Create New Retriever with Texts

retriever = TFIDFRetriever.from_texts(["foo", "bar", "world", "hello", "foo bar"])

Use Retriever

We can now use the retriever!

result = retriever.get_relevant_documents("foo")

result

[Document(page_content='foo', metadata={}),
 Document(page_content='foo bar', metadata={}),
 Document(page_content='hello', metadata={}),
 Document(page_content='world', metadata={})]