新冠肺炎疫情下专业图书馆科研新媒体服务探究——以中国科学院文献情报中心“中国科讯”为例

[目的/意义] 以中国科学院文献情报中心"中国科讯"为例，分析专业图书馆应急科学传播服务的举措及成效，为应急事件下专业图书馆科学传播服务提供借鉴。[方法/过程] 梳理新冠肺炎疫情期间"中国科讯"微信公众号推送的服务内容，从日均用户增长和阅读量分析用户对不同类型内容的关注程度，提出应急科学传播服务的思考。[结果/结论] 疫情期间，"中国科讯"日均用户增长明显，电子资源保障类内容阅读量最高，其次为新冠病毒科技攻关进展相关内容、科普类直播和信息素养教育直播观看人数达到数万人。应急事件下，专业图书馆应加强特色服务的宣传、加强信息甄选和审核机制，重大公共卫生事件下应急科学传播工作预案缺失也需引起重视。     

沉默还是回忆:战后德国记忆文化评析

今天德国秉承的记忆文化,主要是对犹太人大屠杀历史的直面,经历了一个曲折、漫长的过程。初期德国老一代表现出集体沉默,而中期德国年轻一代打破沉默,到后期德国集体树立反省意识。德国对大屠杀的这一反省进程在国内外政治变迁、人口世代更替、国家统一实现这些因素综合的作用之下,而不断地趋向正面。如今德国完成了对大屠杀罪行记忆的恒久化与规范化。     

The Grand Challenges Discourse: Transforming Identity Work in Science and Science Policy

This article analyzes the concept of “grand challenges” as part of a shift in how scientists and policymakers frame and communicate their respective agendas. The history of the grand challenges discourse helps to understand how identity work in science and science policy has been transformed in recent decades. Furthermore, the question is raised whether this discourse is only an indicator, or also a factor in this transformation. Building on conceptual history and historical semantics, the two parts of the article reconstruct two discursive shifts. First, the observation that in scientific communication references to “problems” are increasingly substituted by references to “challenges” indicates a broader cultural trend of how attitudes towards what is problematic have shifted in the last decades. Second, as the grand challenges discourse is rooted in the sphere of sports and competition, it introduces a specific new set of societal values and practices into the spheres of science and technology. The article concludes that this process can be characterized as the sportification of science, which contributes to self-mobilization and, ultimately, to self-optimization of the participating scientists, engineers, and policymakers.     

基于深度学习的中文专利自动分类方法研究

[目的/意义] 面向当前国内专利审查和专利情报分析工作中对于海量专利分类的客观需求，设计了7种基于深度学习的专利自动分类方法，对比各种方法的分类效果，从而助力专利分类效率和效果的提升。[方法/过程] 针对传统机器学习方法存在的缺陷，基于Word2Vec、CNN、RNN、Attention机制等深度学习技术，考虑专利文本语序特征、上下文特征以及分类关键特征，设计Word2Vec+TextCNN、Word2Vec+GRU、Word2Vec+BiGRU、Word2Vec+BiGRU+TextCNN等7种深度学习模型，以中国专利为例，选取IPC主分类号的"部"作为分类依据，对比这7种模型与3种传统分类模型在中文专利分类任务中的效果。[结果/结论] 实证研究效果显示，采用考虑语序特征、上下文特征及强化关键特征的深度学习方法进行中文专利分类具有更优的分类效果。