框架理论视角下人工智能风险的社会建构：基于事件新闻文本挖掘的分析

打开文本图片集

The Social Construction of Al Risks from the Perspective of Framing Theory： An Analysis Based on Event News Text Mining

AbstractThroughtheframing strategies，the mediaconstructesocialmeaningofartificial intelligence （AI）risks，shaping public perceptionand influencing policyresponses.Clarifying the theme characteristics and emotional expressions in AIriskreportingcan provide theoretical supportand empirical evidence for improvingAIrisk governance.This paper， groundedinframing theory，proposesathree-dimensionalanalyticalmodelof“Scope-Perspective-Tone”andemploys naturallanguageprocessing techniquesincludingLDAtopicmodeling，sentimentanalysis，andlinguisticambiguitydetectionto systematicallexamineAIrisk news texts，uncovering themechanismsandevolutionof mediaframing inthis context.Itturnsout theframingofAIriskinthenews hasshiftedfromafocusontechnicalissues—suchasdata privacyand algorithmicbiasinearlieryears—tobroadersociopoliticalconcerns inrecentyears，includingalgorithmicdiscrimination， election interference，andpsychologicalmanipulation.Distincttypesofmediaoutletsdemonstratesignificantdierences inreporting stanceand focal points，reflecting theconstructionof diverse perspectives inriskcommunication.Moreover， the media generalltends toadoptnegative emotional tonesandemploy vague orhedging language，whichamplifies publicanxietyandalertnessregardingAIrisks.Researchshows thatmediaarenotmerelyconduitsofinformationbutactive agents in shaping the social meaning of AI-related risks.

KeywordsAI risk; framing theory;LDA;sentiment analysis;risk evolution

人工智能（ArtificialIntelligence，AI）技术在医疗、教育、金融、交通和司法等领域的广泛应用，不仅显著提升了工作效率，还推动了技术创新[。（剩余18051字）