联系我们

论文查询结果	返回搜索

论文编号：	11090
作者编号：	2320170832
上传时间：	2019/12/6 10:30:19
中文题目：	基于文本挖掘的社会组织政策主题及政策态度演化研究
英文题目：	Research on the Change of Policy Topics and Policy Attitudes of Social Organizations Based on Text Mining
指导老师：	王芳
中文关键字：	社会组织；主题演化；政策态度；文本挖掘；态度特征
英文关键字：	social organization; theme evolution; policy attitude; text mining; attitudinal characteristic
中文摘要：	我国社会组织政策的构建和演变深刻地嵌入了社会组织转型发展的历史过程，但迄今对于政策文本的挖掘和分析较少，对于政策态度变化缺乏量化认知。为此，有必要基于大数据技术对于社会组织政策文本进行深入挖掘，探索社会组织政策的主题和态度演化趋势。本文对1980-2018年社会组织政策文本进行数据挖掘和政策态度分析。首先通过文献计量的方法揭示出社会组织政策的关键词词频、政策数量、发文机关关系等外部特征。接着使用LDA主题模型从预处理后的文本中自动抽取主题，进行聚类，使用多维量表法计算出政策主题间距离进行可视化展示，探究主题强度时间序列演化趋势。然后以评价理论为引导，使用word2vec对社会组织政策文本进行训练，找出相似词语，提高人工扩展态度词典的效率。通过文本态度极性占比的计算建立态度识别模型，对政策文本中蕴含的政府态度进行识别和定量分析。利用此模型，分析社会组织政策文本总体态度极性占比情况、态度主题特征、态度空间特征并进行可视化展示，探讨政策态度在不同时间、不同省份的变化趋势。最后使用内容分析方法，通过政策案例对政策文本进行定性研究，验证政策态度识别模型的有效性。借此理解社会组织发展的历史路径和复杂特征，为社会组织政策的合理构建提供战略性启示。研究发现，社会组织政策态度存在着较为明显的阶段特征，社会生活领域的重大事件会对政策态度产生影响。总体上看，政府对于社会组织的认识趋于深入和理性。政府的社会组织的选择性培育政策反映在政策文本中，不同的主题，政策态度也有所不同。国家部委的方针政策及其反映出的党和中央政府对于社会组织的态度，决定了我国各省市社会组织政策的基调。此外，各省市政策的态度还受到经济发展水平、功能定位、突发事件和地方政府认知理念影响。本文构建的政策文本的态度识别模型将文本挖掘技术应用于态度识别，有助于扩展该领域的研究思路和方法。一方面对社会组织政策的解读与研究提供新的思路和视角，另一方面该模型具有一定的泛化能力，可以为其他领域政策文本的分析提供一些借鉴和参考。
英文摘要：	The construction and evolution of China's social organization policies are deeply embedded in the historical process of social organization transformation and development. However, so far there are few mining and analysis of policy texts, and lack of quantitative recognition of the change of policy attitudes. Therefore, it is necessary to dig deeply into the text of social organization policies based on big data technology, explore the theme evolution and attitude change trend of social organization policies. In this paper, the full text data of social organization policy from 1980 to 2018 are analyzed by data mining. Firstly, the external characteristics of social organization policy, such as keyword frequency, number of policies and the relationship between publishing organs, are revealed by bibliometrics. Then LDA topic model is used to automatically extract topics from pre-processed texts and cluster them. The distance between policy topics is calculated by multi-dimensional scale method for visual display, and the evolutionary trend of theme intensity time series is explored. Then, guided by the evaluation theory, we use Word2vec to train the policy texts of social organizations, find out similar words, and improve the efficiency of manual expansion of attitude dictionary. This paper establishes an attitude recognition model by calculating the polarity proportion of the text attitude, identifies and quantitatively analyses the government attitude contained in the policy text. Using this model, the paper explores the proportion of polarity of attitudes, attitudinal thematic characteristics, attitudinal spatial characteristics and development trends of social organization policy texts, visualizes the changing trends of policy attitudes at different times and in different provinces. Finally, we use content analysis method to conduct qualitative research on policy texts through policy cases to verify the validity of the policy attitude recognition model. In this way, we can understand the historical path and complex characteristics of the development of social organizations, provide strategic enlightenment for the rational construction of social organization policies. It is found that the social organization policy attitude has obvious stage characteristics, and the major events in the field of social life will have an impact on the policy attitude. Generally speaking, the government's understanding of social organizations tends to be in-depth and rational. The selective cultivation policy of the government's social organizations is reflected in the policy text, and the policy attitude varies with different themes. The principles and policies of national ministries and commissions and the attitudes of the party and the central government towards social organizations determine the keynote of social organization policies of all provinces and cities in China. In addition, the policy attitude of provinces and cities is also affected by the level of economic development, functional orientation, emergencies and local government cognitive concept. The attitudinal recognition model of policy text constructed in this paper introduces text mining technology into attitudinal recognition research, which is beneficial to broaden the research ideas and methods in this field. On the one hand, it provides new ideas and perspectives for the interpretation and research of social organization policies. On the other hand, the model has certain generalization ability, and has certain reference value for the analysis of policy texts in other fields.
查看全文：	预览  下载(下载需要进行登录)