In a hypothetical "Contacts" LUIS.AI application with three list entities
ContactType: [Phone, Email]
Country: [UK, Germany, US, Canada, Brazil, Venezuela]
Region: [Europe, North America, South America]
Where the Country entity "US" has synonyms "America, American, USA".
Is it possible to enforce that if an utterance contains the Region "South America", e.g. "What are my phone contacts in South America" that it should (correctly) match the South America "Region" entity, and not (incorrectly) include the US "Country" entity?
I am seeing utterances where the intent is being correctly identified, with the ContactType, and Region; but the Country is also being identified, leading to contextually incorrect responses from the business logic of the app.
In the utterance review of the intent, I see all three entities highlighted in the intent review, but can't figure out how to remove a labelled entity if LUIS decides that it is present.
If it is not possible for LUIS to learn rules such that an entity cannot contain a nested entity, is there a strategy for the business logic to identify that the entity is nested, and should be ignored, other than special casing?
Edit to provide more information:
The defined Entities:
An example of a correctly processed utterance; The intent is correct, the contact method is phone numbers, the contact type is family, and the country is US.
Here is an incorrectly processed utterance; The intent is correct, the contact method and type are correct, but "South America" is incorrectly producing two entitites, a Region (correct) of South America, but also the Country US, which is contextually incorrect.
