Tutorial: Filtering Documents with Metadata


This tutorial uses Haystack 2.0. To learn more, read the Haystack 2.0 announcement or visit the Haystack 2.0 Documentation.

Overview

๐Ÿ“š Useful Documentation: Metadata Filtering

Although new retrieval techniques are great, sometimes you just know that you want to perform search on a specific group of documents in your document store. This can be anything from all the documents that are related to a specific user, or that were published after a certain date and so on. Metadata filtering is very useful in these situations. In this tutorial, we will create a few simple documents containing information about Haystack, where the metadata includes information on what version of Haystack the information relates to. We will then do metadata filtering to make sure we are answering the question based only on information about Haystack 2.0.

Preparing the Colab Environment

Installing Haystack

Install Haystack 2.0 with pip:

%%bash

pip install haystack-ai

Enabling Telemetry

Knowing you’re using this tutorial helps us decide where to invest our efforts to build a better product but you can always opt out by commenting the following line. See Telemetry for more details.

from haystack.telemetry import tutorial_running

tutorial_running(31)

Preparing Documents

First, let’s prepare some documents. Below, we’re manually creating 3 simple documents with meta attached. We’re then writing these documents to an InMemoryDocumentStore, but you can use any of the available document stores instead such as OpenSearch, Chroma, Pinecone and more.. (Note that not all of them have options to store in memory and may require extra setup).

โญ๏ธ For more information on how to write documents into different document stores, you can follow our tutorial on indexing different file types.

from datetime import datetime

from haystack import Document
from haystack.document_stores.in_memory import InMemoryDocumentStore
from haystack.components.retrievers.in_memory import InMemoryBM25Retriever

documents = [
    Document(
        content="Use pip to install a basic version of Haystack's latest release: pip install farm-haystack. All the core Haystack components live in the haystack repo. But there's also the haystack-extras repo which contains components that are not as widely used, and you need to install them separately.",
        meta={"version": 1.15, "date": datetime(2023, 3, 30)},
    ),
    Document(
        content="Use pip to install a basic version of Haystack's latest release: pip install farm-haystack[inference]. All the core Haystack components live in the haystack repo. But there's also the haystack-extras repo which contains components that are not as widely used, and you need to install them separately.",
        meta={"version": 1.22, "date": datetime(2023, 11, 7)},
    ),
    Document(
        content="Use pip to install only the Haystack 2.0 code: pip install haystack-ai. The haystack-ai package is built on the main branch which is an unstable beta version, but it's useful if you want to try the new features as soon as they are merged.",
        meta={"version": 2.0, "date": datetime(2023, 12, 4)},
    ),
]
document_store = InMemoryDocumentStore(bm25_algorithm="BM25Plus")
document_store.write_documents(documents=documents)

Building a Document Search Pipeline

As an example, below we are building a simple document search pipeline that simply has a retriever. However, you can also change this pipeline to do more, such as generating answers to questions or more.

from haystack import Pipeline

pipeline = Pipeline()
pipeline.add_component(instance=InMemoryBM25Retriever(document_store=document_store), name="retriever")

Do Metadata Filtering

Finally, ask a question by filtering the documents to "version" > 1.21.

To see what kind of comparison operators you can use for your metadata, including logical comparistons such as NOT, AND and so on, check out the Metadata Filtering documentation

query = "Haystack installation"
pipeline.run(data={"retriever": {"query": query, "filters": {"field": "meta.version", "operator": ">", "value": 1.21}}})

As a final step, let’s see how we can add logical operators to our filters. This time, we are asking for retrieved documents to be filtered to version > 1.21 AND we’re also asking their date to be later than November 7th 2023.

query = "Haystack installation"
pipeline.run(
    data={
        "retriever": {
            "query": query,
            "filters": {
                "operator": "AND",
                "conditions": [
                    {"field": "meta.version", "operator": ">", "value": 1.21},
                    {"field": "meta.date", "operator": ">", "value": datetime(2023, 11, 7)},
                ],
            },
        }
    }
)

What’s next

๐ŸŽ‰ Congratulations! You’ve filtered retrieved documents with metadata!

If you liked this tutorial, you may also enjoy:

To stay up to date on the latest Haystack developments, you can sign up for our newsletter. Thanks for reading!