Use Machine Learning to Classify Documents

You will use Swagger UI, via all major web browser, to call the Document Classification APIs. Swagger UI allows developers to effortlessly interact and try out every single operation an API exposes for easy consumption. For more information, see Swagger UI.

In the service key you created for Document Classification in the previous tutorial: Create Service Instance for Document Classification with Trial Account, you should find (outside the uaa section of the service key) an entry called url (as highlighted in the image below).

1. To access the Document Classification Swagger UI, add /document-classification/v1 to the url value, paste it in any web browser and press Enter.

   

2. To be able to use the Swagger UI endpoints you need to authorize yourself. In the top right corner, click Authorize.

   

3. Copy the access_token created in the previous tutorial Get OAuth Access Token for Document Classification Using Any Web Browser, add Bearer in front of it, and enter in the Value field.

   Bearer <access_token>
   

   

4. Click Authorize, and then click Close.

   

In this step, you can upload documents and get the service classification results. Please refer to the documentation about the file formats that are supported by the service.

In the Document Classification trial, there is a pre-trained model available for all users. The model predicts whether a document is an invoice, purchase order or payment advice. Feel free to use and upload any of such documents.

As an alternative to uploading your own documents to the service, you can use the following sample files (right click on the link, then click Save link as to download the files locally):

• Invoice (“Beijing Olympics logo” by sha3bi1Beijing)
• Payment Advice (“Logo of Flickr” by topgold)
• Purchase Order (“Logo Guru Shop” by Carlo Armanni)
• Other (“Logo of Twitter” by topgold)

Please note that those files are fake invoices, purchase orders, payment advices and other business documents for demo purposes. Logos were downloaded from search.creativecommons.org, all licensed under CC BY 2.0. Logos were embedded in the documents unaltered. The creators of the logos are denoted in brackets.

In Swagger UI, open the group classification. Then click the second endpoint, called /classification/models/{modelName}/versions/{modelVersion}/documents. The request is marked by POST which indicates that data is sent to the service.

Next, click Try it out to try the request.

Once you click the button, several input fields show up:

1. In the field document, upload the document you want to classify.

2. The field parameters should be empty. In case the field has any pre-filled content, go ahead and delete it.

3. The field modelVersion needs to be filled. As machine learning models can have multiple versions, the version of the model needs to be specified with every request. Write latest into the field so that the latest version of the model is used.

4. The field modelName needs to be specified. Enter DocumentInfoRecord/AUT to use the model that categorizes products.

5. Finally, click Execute to send the document to the service.

Once you have executed the request, you will receive a response from the server. To see the response, scroll down a bit until you find the Response body.

The response includes the current status of the classification process. The status should be RUNNING, which indicates that the classification is in process. Additionally, a documentId is included in the response. This ID is automatically assigned to the document by the server. Copy or make note of the ID, as you will need it in the next step.

To obtain the results of the classification, a different endpoint has to be used.

Open the third endpoint within the group classification, called /classification/models/{modelName}/versions/{modelVersion}/documents/{documentId}. The GET in front indicates that this time data is requested from the server.

Click Try it out to try the request.

Once you click the button, similar input fields, as before, show up:

1. Enter latest in the field modelVersion.
2. Enter DocumentInfoRecord/AUT in the field modelName.
3. In the field documentId, enter the ID you received in the previous step.

Finally, click Execute to request the classification results.

Once you have executed the request, you will receive a response from the server which you find under Response body, as in the previous step. This time, the response includes the classification results for every characteristic by which the document is classified.

The only characteristic in this model is document_type. The possible types of the characteristic are invoice, payment advice and purchase order. The label indicates the type and the associated score represents how confident the service is about the prediction - the higher value the more confident the model is.

In the screenshot below, the sample purchase order was used.

Go ahead and upload other documents to see the outcome. Just note that there is a limited amount of documents that you can classify using a trial account. Find the trial account input limits for Document Classification here.

Congratulations, you have now successfully used our machine learning model to classify business documents, namely invoice, payment advices and purchase orders.