Table of Contents

Training the model for the IVA channel


Only available versions of this content are shown in the dropdown

To enable Pega Intelligent Virtual Assistant™ (IVA) in your application to react to user requests, teach the system to recognize different types of input in a chat conversation. When you train the text analytics model for the IVA, the system learns from training records, improves the artificial intelligence algorithms, and learns to react in the correct way.

For example, after you train the text analytics model for an IVA and a user asks about a car insurance quote in the chatbot, the system detects the correct topic and entities. As a result, the IVA starts a car insurance business case and asks the user several follow-up questions about their car and credit history.
Build an IVA by configuring a Digital Messaging channel. For more information, see Creating a Digital Messaging channel.

To ensure that the IVA learns from the training records and detects the correct topics and entities, apply the training changes to the text analytics model for the system. You apply changes to the model after you enable the training data recording, and then create and review the training records.

There are several guidelines to consider when you train the text analytics model for your chatbot. The approach that you take when training model-based entities is different than when you are building rule-based entities. By observing the guidelines for training text analytics models, you decrease the time it takes to train an IVA by improving the accuracy of the model, so that the system can more efficiently detect correct, relevant information in a conversation. For more information, see Best practices when training model-based entities in an IVA, Best practices when building rule-based entities in an IVA, and Best practices for cleaning up training data in an IVA.

To learn how to improve the user experience in the IVA for Digital Messaging, see Improving the user experience of chatbots.
  1. Enable training of your text analytics model for the IVA.

  2. Improve the IVA model by providing sample training records.

  3. Update the training data by ensuring that the IVA detects the correct information by using text analysis.

    For more information, see Correcting training data in an IVA.
  4. Optional:

    To copy the training data to another IVA, export and import the training records.

    This action makes the IVA easier to maintain and build in production, QA, and development environments. For more information, see Transferring training data to another IVA.
  5. Make the IVA learn from the training data by building a text analytics model.

    This action improves the artificial intelligence of the system. For more information, see Applying changes to a text analytics model for an IVA.
Did you find this content helpful?

Have a question? Get answers now.

Visit the Collaboration Center to ask questions, engage in discussions, share ideas, and help others.