Map a topic to a secondary NLU model

  • Release version: Washingtondc
  • Updated February 1, 2024
  • 1 minute to read

    • Map a Virtual Agent topic to a secondary model in the model group or to a standalone, single-language NLU model for each supported language.

    Before you begin

    If you're using ServiceNow NLU, review NLU model mapping in Virtual Agent Designer for an understanding of NLU model groups.

    If necessary, add secondary languages to the model group in NLU Workbench to create language-specific versions of the model. For more information, see Multilingual model management.

    Role required: virtual_agent_admin or admin

    About this task

    If you select a secondary model for a language that is part of a different model group, the mapping changes for the entire topic. Secondary models must be part of the same model group as the primary language. You can select a different model and intent for a secondary language when it is the only language in the model (and therefore the primary language).
    Figure 1. Updated model group mappings message
    A mapping change alert says, "We updated mappings for these languages: English, French, German, Spanish."
    Figure 2. Topic mapped to a model group with primary and secondary languages
    On the Intent mapping tab, a topic is mapped to a model group with an English primary language model and a German secondary language model.

    For more information, see NLU model mapping in Virtual Agent Designer.

    Procedure

    1. Navigate to All > Conversational Interfaces > Virtual Agent > Designer and open a topic.
    2. Open the Languages tab, and then select the NLU language mapping tab.
      The first column is the NLU mapping that is based on your current session language. For example, if you're using English, the first column is English.
    3. In the Language to edit list, select the language to map.
      The languages listed are based on the NLU languages that you enabled in Virtual Agent settings. After you select a language, a column displays in the Intent mapping section and, if applicable, in the Entity mapping section. For example, if you select Français, a French language column displays.

      The NLU Language Mapping tab displays a column for English and a column for French language mappings. The English model and intent are mapped to a French model and intent.

    4. Update the Intent mapping section.
      1. In the Model row, select an NLU model for the language that you selected.
      2. In the Intent row, select an NLU intent for the language that you selected.
    5. If applicable, update the Entity mapping section.
      For each language, map an entity from the model to an input node in the topic. In a conversation, these nodes are populated automatically with the entity value.
    6. Click Save.

    What to do next

    Test topic and NLU model translations