Use Machine Learning to Enrich Business Data

0 %

Use Machine Learning to Enrich Business Data

Details

Use Machine Learning to Enrich Business Data

Juliana Morais

September 15, 2020

Created by

April 21, 2020

Intermediate

15 min.

Create, update, list and delete enrichment data using the Data API business entity from Document Information Extraction, one of the SAP AI Business Services in SAP Cloud Platform.

You will learn

How to create, update, list and delete enrichment data using the business entity

Prerequisites

After completing the tutorial mission Get Started with Document Information Extraction, you can also use Document Information Extraction to enrich the information extracted from documents with your own master data. You can, for example, match enrichment data entities, such as supplier IDs, with the document Extracted Header Fields, such as sender names.

When enriching data with Document Information Extraction, you use 2 types of entities that you find in business documents. The business entity represents different kinds of organizations with which you deal as a company. It can represent, for example, suppliers and customers. The employee entity represents an employee in the company.

When you finish this tutorial, you will have explored all Data API functionalities to create, update, list and delete enrichment data using the business entity type. See Enrichment Data API documentation.

Step 1: Create client

After completing the prerequisite tutorials Create Service Keys for Document Information Extraction and Get OAuth Access Token for Document Information Extraction Using Any Web Browser, and performing at least step 1 in the tutorial Use Machine Learning to Extract Information from Documents to access and authorize the Document Information Extraction Swagger UI, you need to create a client. This client is used in most of the endpoints to distinguish and separate data.

You can either create a single client or multiple clients in the payload field of the POST /clients endpoint. The clientId values created here will be used in other service endpoints.

Expand the POST /clients endpoint.
Click Try it out.
Enter your clientId and clientName values in the payload field in the format you see in Examples for payload parameter (c_29 and client 29, for example).
Click Execute.

You should receive a response like the following:

CAUTION:

Be aware of the following Document Information Extraction trial account limitations:

Maximum 40 uploaded documents per week

Maximum 1 created clientId

Maximum 10 created enrichment dataIds

Done

Log on to answer question

Step 2: Create enrichment data

View Full Instructions

Use the POST /data/jobs endpoint to add your own master data records to the database to enrich the information extracted from documents.

Expand the POST /data/jobs endpoint.
Click Try it out.

Define the data in the payload field, so that the system knows which extracted field (using, for example, supplier IDs from master data) should be enriched.

{
   "value":[
      {
         "id":"BE0001",
         "name":"Sliced Invoices",
         "accountNumber":"",
         "address1":"123 Somewhere Street Your AZ 12345 123 Somewhere St Melbourne, VIC 3000",
         "address2":"",
         "city":"",
         "countryCode":"",
         "postalCode":"",
         "state":"",
         "email":"",
         "phone":"",
         "bankAccount":"",
         "taxId":""
      },
      {
         "id":"BE0002",
         "name":"Sliced",
         "accountNumber":"",
         "address1":"123 Somewhere Street Your AZ 12345 123 Somewhere St Melbourne, VIC 3000",
         "address2":"",
         "city":"",
         "countryCode":"",
         "postalCode":"",
         "state":"",
         "email":"",
         "phone":"",
         "bankAccount":"",
         "taxId":""
      }
   ]
}

Choose the enrichment data type businessEntity.
Enter your clientId (created in the previous step).
When you choose the enrichment data type business entity, you have the option to choose a subtype (supplier, customer or companyCode). In this example, choose supplier.
Click Execute.

What just happened?

In this example, in the payload field, several master data records (name, ID and address, for example) from 2 different suppliers (Sliced Invoices and Sliced) are provided, so this additional information can be added to the document extracted fields prediction when the information matches.

You should receive a response like the following with status PENDING:

Copy the id from the Response body to see the result of the enrichment data status in the next step.

Which of the following values you can choose in the subtype parameter to create enrichment data when using the business entity? accountNumber, address1 and address2 bankAccount, city and countryCode payload, type and clientId supplier, customer, companyCode

Step 3: See created enrichment data status

Use the GET /data/jobs/{id} endpoint to see the status of the uploaded enrichment data.

Expand the GET /data/jobs/{id} endpoint.
Click Try it out.
Enter the id received in the POST /data/jobs endpoint as the id.
Click Execute.

You should receive a response like the following with status SUCCESS:

Done

Log on to answer question

Step 4: Upload document to get prediction with enrichment data

View Full Instructions

Document Information Extraction uses a globally pre-trained machine learning model that currently obtains better accuracy results with invoices and payment advices in English. The team is working to support additional document types and languages in the near future.

When enrichment data has been uploaded and fits to a certain prediction it is added to the results from the GET /document/jobs/{id} endpoint. To have the enrichment data in the prediction, you need to have the following part in the query of the POST /document/jobs endpoint (it is usually already there by default):

```JSON
"enrichment": {
    "sender": {
      "top": 5,
      "type": "businessEntity",
      "subtype": "supplier"
    },
    "employee": {
      "type": "employee"
    }
  }
```

Do the following:

Expand the POST /document/jobs endpoint.
Click Try it out.
Right click Sample Invoice 1, then click Save link as to download locally the document PDF file for this enrich business data example.

You can also upload and enrich your own document in PDF format with content in headers and tables (an invoice, for example). In this case, make sure the data you define in the payload field, in step 2, matches your document fields.
Upload the document PDF file you want to enrich.

In options, enter the list of fields to be extracted from the uploaded file (documentNumber, taxId, purchaseOrderNumber, shippingAmount, netAmount, senderAddress, senderName, grossAmount, for example), the client id you created in step 1 (c_29, for example), the document type (invoice, for example), receivedDate (2020-02-17, for example), the enrichment data type businessEntity and subtype supplier.

{
   "extraction":{
      "headerFields":[
         "documentNumber",
         "taxId",
         "purchaseOrderNumber",
         "shippingAmount",
         "netAmount",
         "senderAddress",
         "senderName",
         "grossAmount",
         "currencyCode",
         "receiverContact",
         "documentDate",
         "taxAmount",
         "taxRate",
         "receiverName",
         "receiverAddress"
      ],
      "lineItemFields":[
         "description",
         "netAmount",
         "quantity",
         "unitPrice",
         "materialNumber"
      ]
   },
   "clientId":"c_29",
   "documentType":"invoice",
   "receivedDate":"2020-02-17",
   "enrichment":{
      "sender":{
         "top":5,
         "type":"businessEntity",
         "subtype":"supplier"
      },
      "employee":{
         "type":"employee"
      }
   }
}

Click Execute.

This is how the request should look like:

And that’s how the response looks like:

Copy the id from the Response body to get enrichment data prediction in the next step.

Done

Log on to answer question

Step 5: Get enrichment data prediction

When enrichment data has been uploaded and fits to a certain prediction it is added to the results from the GET /document/jobs/{id} endpoint.

Expand the GET /document/jobs/{id} endpoint.
Click Try it out.
Set extractedValues to true to get the extracted values.
Enter the id received in the POST /document/jobs endpoint as the id.
Click Execute.

The endpoint request and response look as follows:

What just happened?

In this example, in the response, one of the extracted fields is the sender name Sliced Invoices. This information is enriched with the supplier ID enrichment data created in step 2. The prediction suggests the supplier ID BE0001 (from sender name Sliced Invoices) with higher probability than the supplier ID BE0002 (from sender name Sliced).

This is an example of a full prediction including the enrichment data part:

```JSON
{
  "status": "DONE",
  "id": "97a96f7c-ea6a-456c-a58c-1cee1121a210",
  "fileName": "sample-invoice-1.pdf",
  "documentType": "invoice",
  "created": "2020-06-15T09:59:04.091971+00:00",
  "finished": "2020-06-15T09:59:29.115803+00:00",
  "country": "XX",
  "extraction": {
    "headerFields": [
      {
        "name": "currencyCode",
        "category": "amounts",
        "value": "USD",
        "type": "string",
        "confidence": 0.919029772281647,
        "page": 1,
        "coordinates": {
          "x": 0,
          "y": 0,
          "w": 0,
          "h": 0
        }
      },
      {
        "name": "documentDate",
        "category": "document",
        "value": "2016-01-25",
        "type": "date",
        "confidence": 0.983220152556896,
        "page": 1,
        "coordinates": {
          "x": 0.759677419354839,
          "y": 0.174458380843786,
          "w": 0.113709677419355,
          "h": 0.00997719498289626
        }
      },
      {
        "name": "documentNumber",
        "category": "document",
        "value": "INV-3337",
        "type": "string",
        "confidence": 0.655578226513333,
        "page": 1,
        "coordinates": {
          "x": 0.760887096774194,
          "y": 0.136830102622577,
          "w": 0.0608870967741936,
          "h": 0.00826681870011403
        }
      },
      {
        "name": "grossAmount",
        "category": "amounts",
        "value": 93.5,
        "type": "number",
        "confidence": 0.656007521080248,
        "page": 1,
        "coordinates": {
          "x": 0.870161290322581,
          "y": 0.500285062713797,
          "w": 0.0447580645161291,
          "h": 0.00940706955530213
        }
      },
      {
        "name": "netAmount",
        "category": "amounts",
        "value": 85,
        "type": "number",
        "confidence": 0.64037944782864,
        "page": 1,
        "coordinates": {
          "x": 0.870161290322581,
          "y": 0.462656784492588,
          "w": 0.0447580645161291,
          "h": 0.00940706955530218
        }
      },
      {
        "name": "paymentTerms",
        "category": "payment",
        "value": "within 30 days from date of invoice.",
        "type": "string",
        "confidence": 0.600183615235057,
        "page": 1,
        "coordinates": {
          "x": 0.192741935483871,
          "y": 0.863169897377423,
          "w": 0.23508064516129,
          "h": 0.0102622576966933
        }
      },
      {
        "name": "purchaseOrderNumber",
        "category": "details",
        "value": "12345",
        "type": "string",
        "confidence": 0,
        "page": 1,
        "coordinates": {
          "x": 0.760887096774194,
          "y": 0.155929304446978,
          "w": 0.0391129032258065,
          "h": 0.00741163055872293
        }
      },
      {
        "name": "receiverAddress",
        "category": "receiver",
        "value": "123 Somewhere St Melbourne, VIC 3000",
        "type": "string",
        "confidence": 0.583759787255599,
        "page": 1,
        "coordinates": {
          "x": 0.0729838709677419,
          "y": 0.303591790193843,
          "w": 0.138306451612903,
          "h": 0.024230330672748
        }
      },
      {
        "name": "receiverName",
        "category": "receiver",
        "value": "Test Business",
        "type": "string",
        "confidence": 0.611305298987362,
        "page": 1,
        "coordinates": {
          "x": 0.0717741935483871,
          "y": 0.288768529076397,
          "w": 0.092741935483871,
          "h": 0.00798175598631695
        }
      },
      {
        "name": "senderAddress",
        "category": "sender",
        "value": "Suite 5A-1204 123 Somewhere Street Your City AZ 12345",
        "type": "string",
        "confidence": 0.60411537696182,
        "page": 1,
        "coordinates": {
          "x": 0.0721774193548387,
          "y": 0.173318129988597,
          "w": 0.150806451612903,
          "h": 0.040193842645382
        }
      },
      {
        "name": "senderName",
        "category": "sender",
        "value": "Sliced Invoices",
        "type": "string",
        "confidence": 0.622503629965442,
        "page": 1,
        "coordinates": {
          "x": 0.0713709677419355,
          "y": 0.157069555302166,
          "w": 0.100403225806452,
          "h": 0.00798175598631698
        }
      },
      {
        "name": "taxRate",
        "category": "amounts",
        "value": 0,
        "type": "number",
        "confidence": 0.818773365020752,
        "page": 1,
        "coordinates": {
          "x": 0.743548393249512,
          "y": 0.420182436704636,
          "w": 0.0411290526390076,
          "h": 0.00769668817520142
        },
        "group": 1
      }
    ],
    "lineItems": [
      [
        {
          "name": "description",
          "category": "details",
          "value": "Web Design",
          "type": "string",
          "confidence": 0.607532449066639,
          "page": 1,
          "coordinates": {
            "x": 0.172983870967742,
            "y": 0.414196123147092,
            "w": 0.0794354838709678,
            "h": 0.0102622576966933
          }
        },
        {
          "name": "quantity",
          "category": "details",
          "value": 1,
          "type": "number",
          "confidence": 0.63710602976027,
          "page": 1,
          "coordinates": {
            "x": 0.102822580645161,
            "y": 0.42018244013683,
            "w": 0.0270161290322581,
            "h": 0.00769669327251998
          }
        },
        {
          "name": "netAmount",
          "category": "amounts",
          "value": 85,
          "type": "number",
          "confidence": 0.633846010222579,
          "page": 1,
          "coordinates": {
            "x": 0.870161290322581,
            "y": 0.419612314709236,
            "w": 0.0447580645161291,
            "h": 0.00940706955530213
          }
        },
        {
          "name": "unitPrice",
          "category": "details",
          "value": 85,
          "type": "number",
          "confidence": 0.635079625881079,
          "page": 1,
          "coordinates": {
            "x": 0.610483870967742,
            "y": 0.419612314709236,
            "w": 0.0447580645161291,
            "h": 0.00940706955530213
          }
        }
      ]
    ]
  },
  "enrichment": {
    "sender": [
      {
        "id": "BE0001",
        "confidence": 0.5625
      },
      {
        "id": "BE0002",
        "confidence": 0.4625
      }
    ],
    "employee": []
  }
}    
```

You have now successfully used the business entity to get enrichment data predictions for the document you uploaded to Document Information Extraction.

Done

Log on to answer question

Step 6: See all enrichment data entries

To see a list of the enrichment data entries you have created:

Expand the GET /data endpoint.
Click Try it out.
Choose the enrichment data type businessEntity.
Enter your clientId.
Choose the enrichment data subtype supplier.
Click Execute.

You should receive a response like the following:

Done

Log on to answer question

Step 7: Delete enrichment data

View Full Instructions

To delete enrichment data which has been uploaded before:

Expand the DELETE /data endpoint.
Click Try it out.
Define the data in the payload field, so that the system knows which data entry (using, for example, the data entry ID) should be deleted.
```
{
   "value":[
      {
         "id":"BE0001"
      }
   ]
}
```
Choose the enrichment data type businessEntity.
Enter your clientId.
Choose the enrichment data subtype supplier.
Click Execute.

You should receive a response like the following:

Done

Log on to answer question

Step 8: Delete client

If you want to delete a client you created in Step 1, use the DELETE /clients endpoint.

Expand the DELETE /clients endpoint.
Click Try it out.
Enter in the payload field the client id or multiple client ids (c_29, for example) you want to delete.
Click Execute.