Use zero-shot large language models on Amazon Bedrock for custom named entity recognition

This article discusses how to use large language models (LLMs) on Amazon Bedrock to perform zero-shot named entity recognition (NER) on custom entity types without the need for entity-specific fine-tuning.

Specifically, the article covers:

• Solution overview involving Amazon Textract, Amazon Comprehend, and Amazon Bedrock
• Extracting context from documents using Amazon Textract
• Truncating context using Amazon Comprehend (optional)
• Extracting entity-value pairs using LLMs on Amazon Bedrock
• An example use case demonstrating the solution
• Conclusion and future work