Skip to main content


         This documentation site is for previous versions. Visit our new documentation site for current releases.      
 

Configuring topic settings in text predictions

Updated on July 5, 2022

Use topic settings to control how text is categorized, depending on the selected level of classification granularity. You can adjust text categorization according to your business needs, for example, to change the granularity of the analysis to document level if you analyze short chat messages.

  1. Open the text prediction:
    1. In the navigation pane of App Studio, click Channels.
    2. In the Current channel interfaces section, click the icon that represents a channel for which you want to configure the text prediction.
    3. On the channel configuration page, click the Behavior tab, and then click Open text prediction.
  2. In the Prediction workspace, click the Settings tab, and scroll down to the Topic settings section.
  3. In the Pega NLP section, configure granularity for Pega NLP models:
    ChoicesActions
    Define sentence-level granularity

    In the Pega NLP section, select Sentence level.

    Sentence-level granularity is optimal for high-precision analysis when you want to identify the top topic for each sentence separately. Use this feature to analyze large units of text, for example, emails or blog entries.

    Define document-level granularity
    1. In the Pega NLP section, select Document level, and then click Configure.
    2. In the Document level granularity window, select one of the following options:
      • To detect only a specified number of topics that received the highest confidence score, select Select an amount of top topics, and then enter the appropriate amount.
      • To limit the number of detected topics to only those above a specific confidence score threshold, select Select topics above a confidence score threshold, and then enter the appropriate threshold.

      Document-level granularity is useful when you want to categorize the text as a whole, with no further breakdown. Use this feature to analyze smaller units of text, for example, chat messages.

    3. Optional: Enable the option to fall back to rule-based topic detection if the specified confidence threshold is not met.
    4. Click Submit.
    Configuring the confidence score threshold and fallback setting
    The document level granularity configuration window
  4. In the External models section, configure granularity for external models:
    • To detect only a specific number of topics that received the highest confidence score, select Select an amount of top topics, and then enter the number of topics to detect.
    • To limit the number of detected topics to only those above a specific confidence score threshold, select Select topics above a confidence score threshold, and then enter a threshold value.
  5. In the Taxonomy section, in the Parent topic field, enter a parent topic for all the topics that you add to this prediction.
    By default, the parent topic is set to action.
  6. Click Save.
  • Previous topic Configuring language settings in text predictions
  • Next topic Configuring sentiment settings in text predictions

Have a question? Get answers now.

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

Did you find this content helpful?

Want to help us improve this content?

We'd prefer it if you saw us at our best.

Pega.com is not optimized for Internet Explorer. For the optimal experience, please use:

Close Deprecation Notice
Contact us