Skip to main content


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

Guidelines for sharing text analytics models across channels

Updated on July 22, 2021

Develop and maintain Pega Intelligent Virtual Assistant™ (IVA) for your application more efficiently by reusing text analytics models in multiple IVAs and email bots. Sharing text analytics models makes channel behavior consistent in all of the IVAs and email bots, and decreases the development time for conversational channels.

Reusing text analytics models is an advanced configuration feature for IVAs. For example, you can share the topic model in multiple text analyzers for IVAs, so that the system consistently detects the correct subject matter in conversation text for each IVA.

Channels in your applications, such as IVAs or email bots, work closely with text analyzers. Pega Platform provides you with several out-of-the-box text analytics models for topic, small talk, and sentiment detection. The models are referenced in a text analyzer rule for each channel. When you create an IVA or an Email channel in your application, the system saves a unique instance of the text analyzer and its related text analytics models in the child hierarchy of the Data-Decision-Request base class. The topic model referenced by the pyInteractionTaxonomy rule is saved in the following classes:

  • Data-Decision-Request-Email-<UNIQUE_ID> class for email bots
  • Data-Decision-Request-MCP-<IVA_TYPE>-<UNIQUE_ID> class for IVAs
Note: You can share and modify text analytics models in Dev Studio.

Guidelines

Follow these guidelines when sharing text analytics models across channels:

  • Create instances of text analytics model rules in the parent Data-Decision-Request base class, instead of within the child hierarchy of this class.
  • Select the rule that you save in the Data-Decision-Request base class for each text analyzer rule of an IVA and email bot that you want to share a model.
  • Ensure that there are no instances of this rule defined at the child class level. Otherwise, during rule resolution the system will give preference to the rule at the child class level instead of the one at the parent class level.
  • If you want to reuse the models for only a subset of the IVAs and email bots, to avoid confusion about which version of the model the system uses, save the model rule at the parent Data-Decision-Request base class level with a different name from the default name. For example, for the topic model, save the pyInteractionTaxonomy rule at the parent class level with another name.
  • The small talk model specified by the pySmallTalk rule for a text analyzer is only available for IVAs. This rule is available in the @baseclass base class, which means that any IVA channel can access it. You can only share this model rule in IVAs, not email bots.
  • Do not share instances of text analyzer rules between multiple IVAs and email bots. You can only share the text analytics models referenced by a text analyzer across different channels.

Example

When you create an Email channel, the system saves an instance of the topic model rule for the email bot in the pyInteractionTaxonomy rule in the Data-Decision-Request-Email-EDVVSXTJ class, under DecisionDecision Data. When you create a Legacy Webchat channel, the system saves an instance of the topic model for the IVA in the pyInteractionTaxonomy rule, in the Data-Decision-Request-MCP-WebChat-KKYYUFTN class, under DecisionDecision Data. To share a text analytics model across multiple channels, for example, the topic model pyInteractionTaxonomy rule, save an instance of this rule with another name in a parent Data-Decision-Request base class, under DecisionDecision Data. To share the topic model, you then select the instance of the pyInteractionTaxonomy rule from this location for each text analyzer rule of an IVA or an email bot.

  • Previous topic Configuring system responses for a conversational channel
  • Next topic Simulating a conversation and building a chatbot

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