This is the new CloudFormation Template Reference Guide. Please update your bookmarks and links. For help getting started with CloudFormation, see the AWS CloudFormation User Guide.
AWS::Bedrock::DataSource ParsingConfiguration
Settings for parsing document contents. If you exclude this field, the default parser converts the contents of each
document into text before splitting it into chunks. Specify the parsing strategy to use in the parsingStrategy field and include the relevant configuration, or omit it to use the Amazon Bedrock default parser. For more information, see Parsing options for your data source.
Note
If you specify BEDROCK_DATA_AUTOMATION or BEDROCK_FOUNDATION_MODEL and it fails to parse a file, the Amazon Bedrock default parser will be used instead.
Syntax
To declare this entity in your CloudFormation template, use the following syntax:
JSON
{ "BedrockDataAutomationConfiguration" :BedrockDataAutomationConfiguration, "BedrockFoundationModelConfiguration" :BedrockFoundationModelConfiguration, "ParsingStrategy" :String}
YAML
BedrockDataAutomationConfiguration:BedrockDataAutomationConfigurationBedrockFoundationModelConfiguration:BedrockFoundationModelConfigurationParsingStrategy:String
Properties
BedrockDataAutomationConfiguration-
If you specify
BEDROCK_DATA_AUTOMATIONas the parsing strategy for ingesting your data source, use this object to modify configurations for using the Amazon Bedrock Data Automation parser.Required: No
Type: BedrockDataAutomationConfiguration
Update requires: Replacement
BedrockFoundationModelConfiguration-
If you specify
BEDROCK_FOUNDATION_MODELas the parsing strategy for ingesting your data source, use this object to modify configurations for using a foundation model to parse documents.Required: No
Type: BedrockFoundationModelConfiguration
Update requires: Replacement
ParsingStrategy-
The parsing strategy for the data source.
For managed knowledge bases, the strategy that you can select depends on the embedding model that your knowledge base uses:
-
If your knowledge base uses a native multimodal embedding model, specify
MULTI_MODAL_EMBEDDINGS. With this strategy, files are sent directly to the embedding model instead of being parsed into text. This is the only strategy that is supported for these knowledge bases. -
Otherwise, specify
SMART_PARSING.
For more information, see Customize ingestion for managed knowledge bases.
Required: Yes
Type: String
Allowed values:
BEDROCK_FOUNDATION_MODEL | BEDROCK_DATA_AUTOMATION | SMART_PARSING | MULTI_MODAL_EMBEDDINGSUpdate requires: Replacement
-
