【摘要】 董事、监事和高级管理人员(DSE)构成的高层管理团队是公司财务违规的主要责任群体.将“知识图谱嵌入”技术应用于DSE履历，综合“(个人)基本特征-治理结构-交叉关联”三维特征，从“整体观”视角构建新的违规倾向指标(PIFT).基于Logit或二元Probit模型进行系列实证检验发现：PIFT越高，公司违规概率越大；与若干基于DSE履历少数特定维度的其他指标相比，PIFT具有最高的重要性.进一步检验表明：非独立董事群体对PIFT有效性起主导作用；同行业违规公司比例越高，PIFT的影响越大，但非独立董事的PIFT不受影响；公司总部所在地法治水平越高，随着PIFT增加，违规(被揭露)的可能性越大，但是否国有股权对PIFT的作用没有显著影响.希望这项工作可以为遵循标准金融实证研究范式，将知识图谱嵌入技术应用于公司金融领域建立一个新的研究起点.更多还原

【Abstract】 The top management team, comprising directors, supervisors, and executives(DSEs), is the primary group responsible for corporate financial misconduct. By applying knowledge graph embedding(KGE) to DSE profiles, this paper constructs a novel fraud propensity indicator(PIFT) that captures semantic similarity across three dimensions-(personal) basic traits, governance structure, and cross ties-from a holistic perspective. Empirical tests using Logit and binary Probit models reveal that Higher PIFTs are associated with a greater likelihood of financial violations, and that, compared to single-dimension indicators, PIFT demonstrates superior predictive power. Further analyses show that non-independent directors play a dominant role in driving PIFT’s effectiveness, that the proportion of fraudulent firms in the same industry strengthens PIFT’s impact, and that higher regional legal enforcement enhances the link between PIFT and fraud risk, while state ownership remains insignificant. This study establishes a new starting point for applying knowledge graph embedding in corporate finance research, following the standard empirical financial research paradigm.更多还原