I have an application built using the latest version of Rasa Open Source which can take user inputs like: I want 3 bananas, an apple and 4 oranges I want 3 bananas, a large apple and 4 medium size oranges I am identifying bananas, apples and oranges under a [fruit] entity, the quantity 3,4,1(default) under a [qty] entity and size medium, large, regular(default) under [size] entity. Thus my nlu file reads like this - I want [3](qty) [bananas](fruit), an [apple](fruit) and [4](qty) [oranges](entity) - I want [3](qty) [bananas](fruit), a [large](size) [apple](fruit) and [4](qty) [medium](size) size [oranges](fruit) The NLU fails to recognize the appropriate quantity and size corresponding to the respective entities. For e.g. in the first case it assigns the quantity 4 to apple instead of oranges and orange is assigned the default value 1. Similarly in the second case, the size large gets applied to bananas and the medium size to apple and regular(default) size to oranges. How do you have the appropriate descriptive entities like size, quantity etc. applied correctly to the respective main entity (in this case, fruit) when the user has a choice to either provide or not provide the descriptive entities?

Accurately identifying size and quantity of individual entities in a multiple entity statement where it is optionally provided

Tanja (Tanja Bunk) June 22, 2020, 11:30am 2

I guess the group label might help: Introducing entity roles and groups

Topic		Replies	Views
Help with intent and entity extraction Rasa Open Source	5	830	February 8, 2021
Same entity types occurring together in a text meeses up the entity and group prediction Rasa Open Source	1	331	June 7, 2021
Entities with multiple roles? Rasa Open Source	0	290	March 22, 2022
Answer doesn't match with entity values Rasa Open Source	7	551	June 28, 2022
Rasa NLU - classification by position Rasa Open Source	2	614	August 1, 2019