Postal mail addresses are widely used in many industries, i.e. transportation, delivery services, invoices, purchase orders, mortgage and e-commerce etc. For those industries, our customers have to extract specific address information from thousands of their documents such as their contact details including mail addresses.
Let’s take the case of the transportation service owner who receives his purchase orders online. From the orders received, he has to extract data manually and send in the delivery addresses to his drivers. They will not only need the contact details of the shipping destination but also the location of the interested property (street, zip code, country). Unfortunately, addresses often come in all sorts of shapes and formats, especially if they have been entered by a person without being verified. When you receive hundreds of purchase orders on a daily basis, it can become time-consuming and difficult to extract address data quickly, especially if you need to integrate those into a CRM or an ERP system.
This is an example of a postal address block often come in one type of mail format in a document, and it shows the components that can be extracted from the block.
Starting from version 22.0.1, ADP provides a new feature to help you easily find and extract your postal mail addresses from different types of documents. You only need a few steps to quickly configure the extraction of postal mail addresses from a document, which are explained in this article.
It is worth noting that this feature currently supports only the extraction of mail addresses in English. If you need to support more languages, please contact your administrator and development team. For more information about the limitations in language / country support, see the Automation Document Processing documentation:
Now let's see how to define and use ADP to extract and parse addresses.
Step 1: Define a new field for postal address
- In the Enrich tab, select Document Types.
- Select the document type that you want to work on, and add a new field.
- Select sys.PostalMailAddress as Field Type.
- Enter a key alias name for this address field, "Shipper" in this case.
- Click Next, then Save.
Optional:
You can choose which address components to be extracted & parsed in the Subfields tab. In this case, we keep them as-is.
Step 2: Annotate the region of the postal address
- Upload a sample document.
- Annotate the key and value for the address field.
- Save the change.
- Mark the document as ready for training.
Step 3: Re-train the ML model
- In Extraction mode, go to the Teach model tab.
- Click Train model in the upper-right corner.
- Wait for the training to successfully complete.
Step 4: Validate the result of the address extraction
- Upload a new document.
- Verify the location of the postal address and the parsed result.
Troubleshooting