摘要：

【目的】探究科技期刊微信公众号中人工智能内容议题的焦点、趋势及盲点,促进科技期刊融合发展。【方法】研究借助Python爬取近5年相关文章,运用BERTopic方法剖析热点及演化趋势,利用Gephi聚类梳理国家人工智能政策体系,对比找出议题设置盲点。【结果】关议题分为技术、应用、产品服务和伦理4个维度;对数据规范、模型基础设施、场景覆盖、市场化产品、权益保护、技术/数据安全以及行业规范等议题的讨论有所欠缺;未来人工智能在技术上聚焦模型数据/训练、生成式大模型;蛋白质分子和地质勘测或成热门应用场景;伦理治理将受热议。【结论】据此提出“技术-应用-产品-风险”四维策略优化框架,为科技期刊深度融合发展提供策略支持。

关键词: 科技期刊, 人工智能, 主题演化, BERTopic, 议题设置

Abstract:

[Purposes] This study aims to explore the focal points, trends, and blind spots of artificial intelligence content topics in the WeChat official accounts of scientific and technological journals, so as to promote the integrated development of scientific journals. [Methods] Python was used to crawl relevant articles in the past five years. The BERTopic method was applied to analyze the hot topics and their evolutionary trends, and the Gephi clustering method was utilized to sort out the national artificial intelligence policy system. By comparing, the blind spots in topic setting were identified. [Findings] The relevant topics can be divided into four dimensions: technology, application, product service, and ethics. There is a lack of discussion on issues such as data specification, model infrastructure, scenario coverage, market-oriented products, rights protection, technology/data security, and industry standards. In the future, artificial intelligence technology will focus on model data/training and generative large-scale models. Protein molecules and coal mining may become popular application scenarios, and ethical governance will be a hot topic. [Conclusions] Based on this, a four-dimensional strategic optimization framework of “technology-application-product-risk” was proposed to provide methodological support for the in-depth integrated development of scientific journals.

Key words: Scientific journals, Artificial intelligence, Topic evolution, BERTopic, Agenda setting