社交媒体上的公众评论分析有助于政府洞察公众对特定政策的看法和具体要求。由于公众的诉求主题和话语表达策略各不相同，政府对它们的关注程度也不尽相同。然而，现有研究很少探讨如何从大量的公众评论中自动识别出政府应更加关注的话语表达逻辑。本研究以社交媒体上关于近期基础教育改革政策的公众评论为例，运用多种文本挖掘工具和技术，对公众的政策感知进行了全面分析，包括诉求主题、情感倾向和话语表达逻辑。据此，本文提出了一种自动识别关键话语逻辑的方法，识别出了13个主要诉求主题，分析了每个主题所蕴含的情感倾向，并归纳了常见的话语逻辑形式、话语系统类别和话语表达策略。结合评论的地域标签信息，本文构建了全国范围内的政策情感热力图。进而对相关政策舆情进行全景描绘，为后续政策的制定与落地和政府部门的舆论引导工作提出建议。

Social media public comment analysis can help the government understand the public's views and specific requirements on specific policies. Due to the different themes and discourse expression strategies of the public's demands, the government's attention to them is also different. However, existing research rarely discusses how to automatically identify the logic of discourse expression that the government should pay more attention to from a large number of public comments. This study takes the public comments on the recent basic education reform policy on social media as an example, and uses a variety of text mining tools and techniques to comprehensively analyze the public's perception of the policy, including demand themes, emotional tendencies, and discourse expression logic. Based on this, this paper proposes a method for automatically identifying key discourse logic, identifying 17 main demand themes, analyzing the emotional tendencies of each theme, and summarizing common discourse logic forms, discourse system categories, and discourse expression strategies. Combining the geographical label information of comments, this article constructs a nationwide policy sentiment heat map. Furthermore, it provides a panoramic depiction of relevant policy public opinions, and offers suggestions for the formulation and implementation of subsequent policies and the guidance of public opinion by government departments.