IBM Watson™ Ideas

Welcome to the IBM Watson™ Ideas Portal


We welcome and appreciate your feedback on IBM Watson™ Products to help make them even better than they are today!


If you are looking for troubleshooting help or wondering how to use our products and services, please check the IBM Watson™ documentation. Please do not use the Ideas Portal for reporting bugs - we ask that you report bugs or issues with the product by contacting IBM support.


Before you submit an idea, please perform a search first as a similar idea may have already been reported in the portal.


If a related idea is not yet listed, please create a new idea and include with it a description which includes expected behavior as well as why having this feature would improve the service and how it would address your use case.

Allow the confidence threshold criteria for disambiguation to be configurable

Currently WA (premium) uses a calculation of confidence within 55% of top intent for disambiguation candidates.  This is too sensitive for some training sets (the range increases with higher confidence levels, triggering disambiguation more often than desired - for example, if top intent comes back with 98% confidence, disambiguation is happening when the second intent is just 54%), would like this to be configurable.

I am working with a customer who cannot make use of this premium feature due to disambiguation triggering too often.

  • Guest
  • Nov 6 2018
  • Future Consideration
Why is it useful?
Who would benefit from this IDEA? As the product owner for a WA implementation, I need to be able to set the sensitivity of the disambiguation feature.
How should it work?
Idea Priority
Priority Justification
Customer Name
Submitting Organization
Submitter Tags
  • Attach files
  • Admin
    Brian Loveys commented
    November 19, 2018 22:00

    Need more data from more clients before moving forward.  Future plans here will learn and adjust the thresholds based on additional usage.

  • Nishi Nair commented
    13 Jun 09:11

    This feature will be really handy based on the criticality of the usecase