[目的/意义]知识付费平台的发展依赖于知识生产者与知识消费者的深度参与，精准识别这两类关键用户对优化平台运营与生态建设具有重要意义。[方法/过程]基于对知识付费平台用户特征的分析，提出了一种融合多维特征的关键用户识别框架。该框架从活跃性、专业性与商业价值三个维度刻画用户行为，并引入BERT情感分析量化用户交互中的情感倾向；进一步结合情感加权的LeaderRank算法计算情感得分，采用熵权法确定指标权重，最终通过加权融合构建综合关键性识别体系，实现对用户的关键性排序。[结果/结论]在喜马拉雅FM平台上的实证研究表明，此方法能有效识别具有高商业价值的头部主播（知识生产者），同时也能捕捉垂直领域专家与情感驱动型长尾用户，识别结果与平台生态结构高度吻合；关键听众（知识消费者）被划分为核心价值型、优势互补型与长尾潜力型三类，体现出多维特征下的差异化行为模式，为平台实施精准分层运营提供了实证依据与方法支持。

[Purpose/significance] The sustainable development of knowledge payment platforms relies heavily on the active participation of both knowledge producers and knowledge consumers. Accurately identifying these two types of key users is crucial for optimizing platform operations and fostering a healthy content ecosystem. [Method/process]Based on the context of knowledge payment platforms, this study proposes a key user identification framework that integrates multi-dimensional features. The framework characterizes user behavior from three dimensions: activity, professionalism, and commercial value. It incorporates BERT-based sentiment analysis to quantify the emotional tendencies in user interactions. Furthermore, a sentiment-weighted LeaderRank algorithm is employed to calculate sentiment scores, and the entropy weight method is used to determine indicator weights. Finally, a comprehensive key user identification system is constructed by integrating these multi-dimensional features, generating a final ranking of user criticality. [Result/conclusion] Empirical research on the Ximalaya FM platform demonstrates that the proposed method effectively identifies top knowledge producers with high commercial value, while also capturing vertical domain experts and sentiment-driven long-tail users. The identification results are highly consistent with the platform′s ecological structure. Key knowledge consumers can be classified into three categories: core-value type, complementary-advantage type, and long-tail potential type, reflecting differentiated behavioral patterns under multi-dimensional features.This provides an empirical basis and methodological support for platforms to implement precise hierarchical user management.