You are here:
Use Queries to Extract Information
Enable Amazon Textract Queries to extract information by using natural language queries.
Required Editions
| Available in: Lightning Experience |
| Available in: Automotive Cloud, Consumer Goods Cloud, Education Cloud, Financial Services Cloud, Health Cloud, Manufacturing Cloud, Media Cloud, Net Zero Cloud, Nonprofit Cloud, Public Sector Solutions. View product and edition availability. |
| Intelligent Document Reader is available with the Intelligent Document Reader add-on license. |
| User Permissions Needed | |
|---|---|
| To create document types: | Customize Application |
Note The Maximum Pages and Confidence Score Threshold fields in the global content extraction
settings are available only in Health Cloud.
- From Setup, in the Quick Find box, enter Intelligent Document Reader, and then select Intelligent Document Reader.
- In the Global Content Extraction Settings section, click Edit Settings.
-
Enter the maximum number of consecutive pages from the first page of a document that you
want to send for content extraction at a time.
The default value for Maximum Pages is 5. If a user selects pages that are already scanned, the pages aren’t counted against the limit.
- Enable Amazon Textract Queries to extract information using natural language queries, with a limit of 200 characters.
- Select the Document Type tab.
- Click New Document Type.
- Enter a name, select the Form Type, and add a description for the document type.
-
In the Queries section, add an entry containing an alias and a natural language query. The
alias serves as a descriptive label displayed in the template, while the query defines the
information you want to extract from documents.
Each document type supports up to 15 unique queries.
-
Save your changes.
The Edit Global Content Extraction Settings page opens.
-
Click Save.
Tip- For extracting information from a document, the defined queries for that document type and other enabled Textract APIs (such as Forms and Analyze ID) are used.
- When you create templates, make sure that all queries are unique because repeating queries causes failures during the extraction process.
- When you create a template by using a document type with associated queries, those queries are run during data extraction and show their corresponding aliases as document fields.
- Queries that fail to identify any values within the document aren't shown as fields in your template.
- After a successful query execution, bounding boxes highlight the extracted values within the document, rather than the associated query labels.
Hjalp denne artikkelen med å løse problemet ditt?
La oss få vite det slik at vi kan forbedre!
