KdConv Chinese Multi-domain Knowledge-driven Conversational Dataset

KdConv Full name Chinese multi-domain Knowledge-driven Conversation dataset, is a Chinese multi-domain knowledge-driven dialogue dataset that builds topics in multi-round dialogues on a knowledge graph. KdConv containsThree areas (movies, music and travel)There are 4.5K dialogues with an average of 19.0 utterances.
This corpus is a Chinese multi-domain dataset for multi-turn knowledge-driven dialogues. It is suitable for modeling knowledge interactions in multi-turn human dialogues, includingKnowledge planning, knowledge base, knowledge adaptationwait.