基于BERTopic的人工智能应用场景主题建模研究

打开文本图片集

中图分类号：G203 文献标识码：A DOI：10.11968/tsyqb.1003-6938.2025032

Study on Topic Modeling of Artificial Intelligence Application Scenario Based on BERTopic

AbstractAgainst the backdropof China's vigorous promotionofAI（Artificial Intellgence）application scenarios，this study employsBERTopic toexaminetopicpatterns inAdeploymentcontexts.Initialy3，524newsarticles werecollectedfromThePaper （Pengpai News）and preprocessed foranalysis.Fortopic modeling，theConan-embedding-vl pretrainedlargemodelwasutilizedfortextembedding，folowedbydimensionalityreductionviaUMAP，clusteringthrough HDBSCAN，and topicrepresentation using c-TF-IDF.Topic keywords were furtherrefinedthrough KeyBERT-based optimizationtechniques.Inopicalysis，eyworddistributions wereexaminedacrosslldomains：technological&D， culturaldigitization，regionaleconomiccollaboration，economicdevelopment，financialinnovation，capitalmarkets， healthcare/elderlycare，policycoordination，low-altitudeeconmy，urbandevelopment，ndnewsdissemination.imilarityanalysisrevealed strong inter-topiccorrelations：urbandevelopmentdemonstratedhighsimilarity witheconomicdevelopment，financial innovation，andpolicycoordination;whileeconomicdevelopmentshowed pronounced alignmentwith financialinovationandpolicycoordination.Hierarchical clusteringanddocumentdistributionanalysis indicated varyingdegreesofcross-domain integrationbetween technologicalR&D，culturaldigitization，andpolicycoordination with othertopicareas.Thisresearch，toacrtainextent，elucidatesthecurrentlandscape，latentdemands，andinterconected elements of disruptive applications ofAI.

Key words BERTopic;artificial inteligence;application scenarios; topic modeling; disruptive application

由国家网信办、国家发改委、教育部、科技部等七部门审议通过的《生成式人工智能服务管理暂行办法》自2023年8月15日起施行，“鼓励生成式人工智能技术在各行业、各领域的创新应用，生成积极健康、向上向善的优质内容，探索优化应用场景，构建应用生态体系”]。（剩余12633字）