Abstract: To identify users' recommendation intent in the conversational context, this paper proposes a knowledge-enhaced soft prompt-tuning method for conversational recommendation intent identification, which introduces external knowledge with specially designed strategies to construct a verbalizer. Specifically, considering the special characteristics including context dependency, noise interference, and dynamic intention, we design four different strategies to expand the label word space for modifying soft prompts, and the integration of these strategies is used as the final verbalizer. Experimental results on three datasets demonstrate that our proposed method outperforms existing advanced approaches, including LLMs, by achieving an accuracy of 88% and an improvement of 4% in F₁.