Configuring Classification

I. Creating a Classification Model

The Classification Model is the training data the system uses to classify your documents. You can create more than one Classification model, where each model is responsible for classifying several types of documents, e.g. you can have one model for classifying invoices and POs and another model for classifying job applications and ID cards.

To create a Classification model you need a pre-labeled set of training data, e.g. one folder with invoices and one folder with purchase orders.

Important

The names of the sub-folders containing each class of document samples should match the names of your document types as defined in the Admin panel.

  1. Smart SmartSoft Classifier, and click Classifier ‣ Document Classifier ‣ Train New Classifier.
../_images/classifier-new-model.png

  1. Select a destination folder where your new Classification model will be saved.
  2. Select a primary and secondary languages for the new Classifier model (the two languages can be the same).
  3. Select the folder where your sample documents are stored. Training documents must be separated by document classes into their respective folders. Each folder must be named after the document class it represents and they must match exactly.
  4. Use the UploadModel tool to upload the Classification model to the Application server. Type the Application server’s IP and the Application server’s port. Click Upload and select the model’s folder. Type in a name and a description for the model.

II. Creating relevant document types and batch types

1. Create one or more document types.

  1. In the Admin panel click Create Document type. You can leave the Form (FTF) name empty if you are only going to classify documents and not capture data. You can add a Form name later.

Important

The name of the document type should match a document class in your classification model. The match must be exact since this is case sensitive.

2. Create a batch type

  1. In the Admin panel click Create Batch Type. Details here. What is a batch type?
  2. Select the Document Types that will be part of that Batch Type.
  3. Select the Classification Model that will be used to classify this Batch Type.
  4. Set the Classification mode to Semantic.

III. Enable the Classify and Class Verify Steps in the Workflow

Start SQL Server Management Studio, and in the UserRole table, in the application database, set the status of the roles Classify and optionally ClassVerify to true. This will change the workflow by adding a classification and a classification verification step in the process.

Note

This operation can be done in the Admin Panel under Workflow configuration.


You are now ready to start Classifying documents