ChatGPT cannot clearly differentiate between class and instance in the given natural language question

[candlecao]

(for ChineseTraditionalKnowledge as a case)
There is a natural language question "鼓吹乐这个音乐类型(乐种)主要分布在什么地域？"(translated as "In which regions is the MusicType of WindAndDrumMusic mainly distributed?") for testing.

From the perspective of musicology, this question is clear and professional. However, from the perspective of entity extraction, it's ambiguous.
This ambiguity arises because 鼓吹乐(WindAndDrumMusic) is actually a class, which is a subclass of "音乐类型(乐种)"(MusicType). Consequently, the potential meaning of the question is: In which regions are those instances classifies as 鼓吹乐(WindAndDrumMusic) mainly/respectively distributed?

However, chatGPT tends to interpret 鼓吹乐(WindAndDrumMusic) as an instance, which leads to a generation of an incorrect SPARQL.

[candlecao]

To solve this question, there are several solutions. For example,

1. I changed the way of asking question: 鼓吹乐类的实例主要分布在什么地域？（translated as "In which regions are the instances of WindAndDrumMusic mainly distributed?"）

It worked

3. Create exactly the same instance with the class, in which case, they can have the same URI