Genpact Cora Knowledge Center


Configure the Text Classification Activity



Text Classification is an Artificial Intelligence activity to add text mining functionality to your workflow instance. It helps classify incoming data per selected models and topics.

A text model, searches for related words and returns their count that users can display and use for different purposes.

You can add text models and topics in the Administration site, which can be selected in this activity.

Use Cases

  • Tagging emails: Text classification functionality is applied to the incoming emails to identify customer satisfaction through email tagging. A text model is created with three topics, Satisfied, Neutral, and Not Satisfied with related words. The appearance of related words in the email subject and body identifies the topics and tags the email to it. These email tags can be used to analyze customer satisfaction. 
  • Request classification and task assignment: Text classification activity picks topics from the received requests, classifies the text, and feeds the result to an assign activity that assigns these requests to relevant teams for processing.

Watch a demonstration of how to configure Text Classification.

Add, edit, or delete text model

In the Administration site, go to Administration > Global Settings > Text Classification Models.

  1. To add a new model, click +Add New Record.
  2. Add model name and topics.
  3. To add a new topic, click +Add Topic.
  4. Add topic name and related words.
  5. Click OK.
  6. Click Add.

The models and topics added here are displayed in the model list for selection while configuring the Text Classification activity.

To edit a model, click the edit icon for a model from the list. Make changes, and click Update.
To delete a model, click the delete icon X for a model from the list.

Configure the Text Classification activity

  1. Add the Text Classification activity to your workflow.
    1. In the Text Classification Properties screen, enter a name for the activity, and then click Next.
  2. Select Text Classification Model from the list, and then click Next.
    You can also add, edit, or delete a model from here. Click New..., Edit..., and Delete respectively.
  3. In the Model Details screen, 
    1. Enter a model name.
    2. Click +Add Topic, to add a new topic to the model.
    3. In the Topic Details screen, enter a topic name.
      You need to enter a unique topic name.
    4. Add related words. Use comma (,) separated values to enter multiple related words.
    5. Click OK.
  4. In the Model Details screen, click OK.
  5. On the Text Selection screen, 
    1. Click Text to classify ellipsis to add an expression that will result in data on which classification will be done.
    2. Add a maximum number of results to be returned after classification.
      The text classification can return many results, but only the set maximum number of results are displayed.
  6. Click Finish.

At present, only English language is supported for text classification. English stopwords are not classified with this functionality.
For a list of stopwords, see this page.

Want to learn by doing? 
Check out this hands-on exercise.
Note that only internal users can access the link.