Node Details

  • Name: awsBedrockKBRetriever

  • Type: AWSBedrockKBRetriever

  • Version: 1.0

  • Category: Retrievers

Base Classes

  • AWSBedrockKBRetriever

  • BaseRetriever

Credentials

The node uses AWS API credentials, which can be optionally provided:

  • Credential Name: awsApi

  • Fields:

    • AWS Key

    • AWS Secret

    • AWS Session Token (optional)

Input Parameters

  1. Region (required)

    • Type: Async Options

    • Default: us-east-1

    • Description: AWS region where the Knowledge Base is located

  2. Knowledge Base ID (required)

    • Type: String

    • Description: Unique identifier for the AWS Bedrock Knowledge Base

  3. Query (optional)

    • Type: String

    • Description: Query to retrieve documents from the retriever. If not specified, the user’s question will be used

    • Accepts variables

  4. TopK (optional)

    • Type: Number

    • Default: 5

    • Description: Number of chunks to retrieve

  5. SearchType (optional)

    • Type: Options

    • Options: HYBRID, SEMANTIC

    • Description: Knowledge Base search type. If not specified, the default will be used

  6. Filter (optional)

    • Type: String

    • Description: Knowledge Base retrieval filter. Requires specific syntax as per AWS documentation

Functionality

The node initializes an AmazonKnowledgeBaseRetriever with the provided parameters. It uses the AWS SDK to connect to the Bedrock Knowledge Base service and retrieve relevant information based on the input query and other specified parameters.

Use Cases

  • Question-answering systems that need to retrieve relevant information from large knowledge bases

  • Information retrieval systems that require access to specific AWS Bedrock Knowledge Bases

  • Applications that need to integrate AWS Bedrock’s Knowledge Base capabilities for enhanced search and retrieval functionalities

Notes

  • The node supports both HYBRID and SEMANTIC search types, allowing for flexible querying based on the use case

  • Filters can be applied to refine the search results, but require adherence to AWS documentation for proper syntax

  • The number of retrieved chunks can be customized using the TopK parameter

  • AWS