AWS Bedrock to split document based on document type

Yes, Amazon Bedrock Data Automation can help you split a single PDF document containing multiple document types and save them as individual files based on their types. Here's how you can approach this using Bedrock Data Automation:

1. Create a Data Automation project: First, you'll need to set up a project in Amazon Bedrock Data Automation. This project will define how your documents should be processed.

2. Configure document splitting: In your project configuration, enable the document splitter feature. This allows Bedrock Data Automation to scan the file and split it into individual documents based on context.

3. Define multiple blueprints: Create custom blueprints for each document type you expect (birth certificate, payslip, and contract). You can have up to 40 document blueprints attached to a single project.

4. Automatic document matching: When you process your PDF file, Bedrock Data Automation will automatically match each split document to the appropriate blueprint based on its content and structure.

5. Custom output generation: Each blueprint can define custom extraction logic and output format for its specific document type.

6. Processing the file: Use the InvokeDataAutomationAsync API to start processing your PDF file asynchronously.

7. Retrieve results: Once processing is complete, you can retrieve the results, which will include the split documents classified by type.

This approach allows Bedrock Data Automation to intelligently split your multi-document PDF, classify each section, and process it according to its specific document type. The system can handle varying page lengths for different document types, making it suitable for your use case with birth certificates, payslips, and contracts of different lengths.

Keep in mind that the accuracy of document splitting and classification depends on the quality of your blueprints and the consistency of your document formats. You may need to refine your blueprints over time to improve accuracy.
Sources
Guidance for Multimodal Data Processing Using Amazon Bedrock Data Automation
Bedrock Data Automation projects - Amazon Bedrock
Simplify multimodal generative AI with Amazon Bedrock Data Automation | AWS Machine Learning Blog

By now Data Automation is GA in us-east-1 and us-west-2. See here for examples showing how to use document splitting with bedrock data automation: https://github.com/aws-samples/sample-document-processing-with-amazon-bedrock-data-automation