社区是最基本的社会治理单元,社会风险源头开始呈现出高度不确定性,传统的治理理念不能满足日益变化的现实需要。韧性治理是新时代理论和实践领域更具创新性和适应性的治理理念,在数字时代背景下,促进构建以数字治理为支撑,以韧性社区为目标的社区治理体系,是推动国家治理能力和治理体系现代化的必然选择。数字技术通过自有的信息收集、信息归纳功能,借助信息平台,能够实现对突发事件的及时预警。目前数字化韧性社区治理过程中存在着治理主体不明确、居民参与度低、数字技术成本与社会分化以及过度注重技术层面而忽略了人文关怀等问题。对此,构建有人文情怀的数字化韧性社区,应该进一步推进党建引领、完善数字平台管理与监督、在社区治理中关注人的需要,提高民众满足感和参与感,从而提升社区制度韧性、技术韧性和文化韧性。The community is the most basic social governance unit, and the source of social risks has begun to show a high degree of uncertainty. The traditional governance concept can not meet the changing needs of the reality. Resilient governance is a more innovative and adaptive governance concept in the field of theory and practice in the new era. Under the background of the digital era, promoting the construction of a community governance system with digital governance as the support and resilient community as the goal is an inevitable choice to promote the modernization of national governance capacity and Governance system. Digital technology through its own information collection, information induction function, with the help of information platform, can realize the timely early warning of emergencies. At present, there are some problems in the process of digital resilient community governance, such as unclear governance subjects, low residents’ participation, digital technology costs and social differentiation, and excessive attention to the technical level while ignoring humanistic care. In this regard, to build a digital resilient community with humanistic feelings, we should further promote the guidance of Party building, improve the management and supervision of digital platforms, pay attention to people’s needs in community governance, and improve the people’s sense of satisfaction and participation, so as to enhance the resilience of community system, technology and culture.
[目的] 针对已有文献对谣言语言风格特征和部分真实的双面健康信息关注较少的研究不足，提出了一个考虑语言风格特征的多模态在线健康谣言检测框架（A multimodal wide and deep approach for online health rumor detection considerin...
详细信息
[目的] 针对已有文献对谣言语言风格特征和部分真实的双面健康信息关注较少的研究不足，提出了一个考虑语言风格特征的多模态在线健康谣言检测框架（A multimodal wide and deep approach for online health rumor detection considering language style，MWDLS）。[方法] MWDLS利用亚里士多德修辞理论来提取诉诸情感、诉诸逻辑和诉诸人格的说服语言风格特征，然后基于双向跨模态交互融合策略和门控机制来实现浅层语言风格特征和深层语言内容特征的联合表征学习和分类预测。[结果] 基于一个微博场景真实数据集的一系列对比实验与消融实验发现，MWDLS的平均F1值在目标任务上比基线模型提高了1.75% ～ 11.98%，表明了MWDLS在健康谣言检测任务上的优越性。[局限] 随着大语言模型性能的不断增强，未来可将本文算法框架与大语言模型进行结合，以探索新的性能提升方向。[结论] 本文提出了一个融合语言风格特征和语言内容特征的多模态健康谣言检测框架，并基于真实社交媒体数据集验证了该模型在健康谣言检测任务上的有效性，具有重要的理论与实践意义。
暂无评论