How to handle long documents?

[kynthesis]

Hi everyone. I'm a newcomer who just started learning about Haystack.
I have some very long .docx documents (about 50K to 500K characters).
And I want to develop a QA system that can give answers from these documents.
I would like to ask for some best practices for this situation. Can someone give me a hint?

[anakin87]

Hello @kynthesis...

In Haystack, (long) documents are usually split into smaller chunks, using the PreProcessor node.

Detailed information can be found here:

• Preprocessing tutorial
• Useful tips about Document length

[kynthesis]

Thank you!!