The problem about unname_entity

[fangwanchuan]

{"question": "where all did picasso live?", "results": "{Yes}. Based on the given knowledge triplets, Pablo Picasso lived in France and M\u00e1laga. The triplets indicate multiple places where Picasso lived, linking unnamed entities to specific locations, and these are sufficient to answer the question. Therefore, the answer is {France, M\u00e1laga}.", "reasoning_chains": [[[["Pablo Picasso", "people.person.nationality", "Spain"], ["Pablo Picasso", "people.person.places_lived", "UnName_Entity"], ["Pablo Picasso", "people.person.places_lived", "UnName_Entity"]]], [[["UnName_Entity", "people.place_lived.location", "France"], ["UnName_Entity", "people.place_lived.location", "M\u00e1laga"], ["UnName_Entity", "people.place_lived.person", "Pablo Picasso"]]]]} Can you hlep me?

[thelastNYF]

It seems an issue about CVT node, some researches consider it and treat it specifically. You can see the paper "Interactive-KBQA: Multi-Turn Interactions for Knowledge Base Question Answering with Large Language Models" in ACL 2024 for more details.