数字乡村政策的主题挖掘及注意力配置——基于LDA2Vec和ATM模型的分析

Topic Mining and Attention Allocation in Digital Rural Policies --An Analysis Based on LDA2Vec and ATM Models

探索数字乡村政策细分领域及注意力配置规律,有助于理解政府推动数字乡村建设的工作重心和思路,进而为促进数字乡村长效发展提供助益。本文基于中央和省级政府颁布的56份数字乡村政策文本,运用LDA模型识别数字乡村政策主题,进而通过构建融合LDA主题模型与Word2Vec的LDA2Vec模型来分析政策注意力配置及其变化态势,并运用ATM模型分析政策注意力配置的主体差异。发现数字乡村政策主题覆盖农业农村大数据开发应用等六个领域;政策注意力配置显著聚焦于数字技术在农业领域的推广和应用且呈现持续上升态势,技术赋能乡村治理与社会效益提升的注意力配置则较为有限;中央政府在数字乡村各领域的注意力配置较为均衡,而省级政府则具有明显的“技术”倾向;不同党政部门的注意力配置呈现出差异化协同的特征。

To explore the subdomains and attention allocation patterns within digital rural policies,this study aims to aid in understanding government efforts to promote digital rural development.Such understanding can further contribute to facilitating sustainable progress in digital rural areas.Based on 56 policy texts issued by central and provincial governments regarding digital rural policies,we employed the Latent Dirichlet Allocation(LDA)model to identify digital rural policy topics.Furthermore,we constructed an integrated LDA2Vec model,combining LDA topic modeling with Word2Vec,to analyze attention allocation dynamics.Additionally,we utilized the Attention Topic Model(ATM)to investigate differences in attention allocation across policy subjects.The digital rural policy topics cover six areas,including agricultural and rural big data applications.Attention allocation significantly focuses on the promotion and application of digital technologies in agriculture,showing a consistent upward trend.However,attention to technology-enabled rural governance and social benefits remains relatively limited.While the central government maintains balanced attention across various digital rural domains,provincial governments exhibit a distinct"technology"bias.Different party and government departments demonstrate distinctive collaborative attention patterns.