# Use the Extraction Service

> Extraction Service API usage for Document Understanding, covering how to select extractors based on classification results and call the extraction endpoint.

Extract data from the document using the same `documentId` and the `extractorId`. The `extractorID` is selected based on the classification result. Select a suitable extractor based on the `documentTypeId` returned by the classification. You can discover classifiers either by calling the Discover Classifiers routes, or by calling the Get Project By Id route. [Here](https://docs.uipath.com/document-understanding/automation-suite/2024.10/api-guide/use-the-discovery-service) are more details about how to use the Discovery Service for retrieving information about the project's available classifiers and extractors.

The POST call for Extraction can be done for synchronous or asynchronous consumption.

Each action requires one POST and multiple GET calls (due to polling), meaning that if you want to validate an extraction result, then you need to do a POST and multiple GET calls.

:::note
Extraction data is soft deleted after one day.
:::

The following examples are using a POST call for asynchronous consumption.

Depending on the operation, Document Understanding API calls use the following classes:

* [Document Class](https://docs.uipath.com/activities/other/latest/document-understanding/document-class) for digitized documents.
* [Classification Result](https://docs.uipath.com/activities/other/latest/document-understanding/classification-result-class) for classification results.