A decision-tree-guided LLM pipeline generates natural shopping dialogues and a 3,600-conversation dataset, but the downstream evaluation is a closed loop on the same generated data.
Title resolution pending
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.CL 1years
2025 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
Wizard of Shopping: Target-Oriented E-commerce Dialogue Generation with Decision Tree Branching
A decision-tree-guided LLM pipeline generates natural shopping dialogues and a 3,600-conversation dataset, but the downstream evaluation is a closed loop on the same generated data.