| 孟志华,祁心雨.省级政府债务风险管控政策的主题挖掘——基于BERTopic模型的文本分析[J].情报工程,2026,(2):087-098 |
| 省级政府债务风险管控政策的主题挖掘——基于BERTopic模型的文本分析 |
| Topic Mining of Provincial Government Debt Risk Control Policies: Text Analysis Based on BERTopic Topic |
| |
| DOI： |
| 中文关键词: 债务风险；BERTopic 模型；政策文本挖掘；主题聚类；风险管控 |
| 英文关键词: Debt Risk; BERTopic Model; Policy Document Mining; Theme Clustering; Risk Management and Control |
| 基金项目:甘肃省哲学社会科学规划项目“化债背景下的特许经营项目协同治理审计机制优化研究”（2024YB083）。 |
| 作者 | 单位 | | 孟志华 | 兰州财经大学会计学院 | | 祁心雨 | 兰州财经大学会计学院 |
|
| 摘要点击次数: 28 |
| 全文下载次数: 30 |
| 中文摘要: |
| [目的/意义]省级政府债务风险管控政策对维护地方财政稳定和经济健康发展至关重要。通过文本分析挖掘相关主题，为优化政策提供依据，助力提升省级政府债务风险管理能力。[方法/过程] 运用BERTopic 模型，对省级政府债务风险管控文本进行分析。首先利用该模型的语义理解能力，对文本进行主题挖掘，进而识别出不同主题及其特征词。其次，基于主题层次聚类法，对挖掘出的主题进行层次聚类分析，明确主题间的相似性与关联性，解析出宏观的研究方向。[结果/结论] 应用BERTopic 主题模型识别出8 个主题及其高频关键词，并通过层次聚类整合揭示了省级政府债务风险管控的三大核心方向。结果表明面向省级政府债务风险管控的非结构化政策文本，BERTopic 模型能够更精准地捕捉债务政策文本中的深层语义结构，提升对债务风险管控文本的核心主题挖掘精度，比传统方法更高效、更深入地挖掘出核心政策主题。 |
| 英文摘要: |
| [Objective/Significance] Provincial debt-risk control policies are crucial for mainta-ining local fiscal stability and sustainable economic growth. This study extracts salient the-mes from these policies through text analysis, aiming to inform policy refinement and enhance provincial debt-risk management. [Methods/Processes] We apply the BERTopic model to a corpus of provincial debt-risk control documents. First, use the semantic understanding capability of the model to perform topic mining on the text, thereby identifying different topics and their characteristic words. Next, based on the hierarchical clustering method of topics, perform hierarchical clustering analysis on the mined topics to clarify the similarity and correlation between topics,and to interpret the macro research directions. [Results/Conclusions] The application of the BERTopic model identified 8 topics along with their high-frequency keywords, and through hierarchical clustering integr-ation, three core directions in provincial government debt risk management and control w-ere revealed. Results indicate that when applied to unstructured policy texts on provincial government debt-risk governance, the BERTopic model captures the deep semantic structure-s of debt-policy discourse more accurately than traditional approaches, thereby enhancing t-he precision of core-topic extraction and enabling more efficient and incisive identification of key policy topics. |
|
查看全文
查看/发表评论 下载PDF阅读器 |
| 关闭 |
