Ignore utterances

[ayman_h]

Hi,

Is it possible to ignore specific utterances? We don't want the Virtual agent to search for topics when a user replies back with a 'Hi' or 'Hello'. We could create a custom topic that redirects to the Greeting topic as a failsafe but checking to see if there is a better out-of-the-box solution.

[Ratnakar7 · Accepted solution]

Hi @ayman_h ,

Yes, you can use the Irrelevance detection feature, which improves the prediction accuracy of NLU models by training them to ignore certain utterances. These utterances from your users may not apply to any intent, so should not get any prediction.

Thanks,

Ratnakar

[ayman_h]

Thanks @Ratnakar7 for your quick response. I can't see the 'Irrelevance detection' when I navigate to the NLU workbench.

This is what I see

[Ratnakar7]

Hi @ayman_h ,

Use the nlu_admin or admin role to access Irrelevance Detection.

Irrelevance Detection is available for Virtual Agent models only.

1. Navigate to All > NLU Workbench > Models. The Virtual Agent tab opens by default.
2. Scroll down the list of Virtual Agent models to the Boost your model performance section.
3. Scroll horizontally to locate the card Keep chats focused, and select its button Go to irrelevance detection.

OR you can try with below URL:
 <instance-name>.servicenow.com/now/nlu-workbench/irrelevant-utterances.

Thanks,

Ratnakar

[ayman_h]

Thanks, found out the problem was the 'NLU Workbench - Advanced Features' plugin (sn_nlu_workbench) needs to be at least v7.0.6 or above to activate the Irrelevance detection feature and the new NLU Workbench