運用文字探勘分析1979年至2022年中國政府工作報告

Abstract

中國經過改革開放後躍居世界強權之一，由於經濟上對外開放而政治上對內封閉的原因，其政府任何對外的官方發布都極為重要，中國政府工作報告是中國官方對外闡述未來政策走向的重要管道，本文以較嶄新的研究方法，以Python這個程式語言架構所需的工具如詞頻分析、關鍵字分析、文字雲、LDA、相似度分析結合資料視覺化分析歷經鄧小平、江澤民、胡錦濤、習近平四位為最高領導人時期的政府工作報告，探討歷經不同領導人時期歷任總理政策制定上的差異，並歸納出其施政的重點。 研究結果發現，工作報告組成可歸為八大類成分: 「經濟發展」、「社會發展」、「國際關係」、「政府改革」、「政治思想」、「港澳台」、「國防」、「其他」，鄧小平時期的政府工作報告著重在「經濟發展」、「國際關係」，而江澤民、胡錦濤、習近平時期則著眼於「經濟發展」、「社會發展」；政府工作報告年與年之間的相似度數值也隨著國際情勢波動減緩保持高達七成左右的平均值，說明中國已經步入了穩定的發展中國家階段；除此之外，觀察主題占比中「政治思想」、「國防」的主題分類站比變動，也可發現不同領導人的權力高低差別，這顯示了文字探勘為研究方法的有效性。

After reforming and opening up policy, China has become one of the most powerful country. Due to the economic openness but political block out situation ,any statement make by the Chinese government is undoubtedly important. The Chinese government work report is the official statement of the future policy direction of China. This thesis uses a relatively new research method, using the tools required for the Python programming language framework, such as word frequency analysis, keyword analysis, word cloud, LDA, similarity analysis, and data visualization analysis to analyze the differences in the policy formulation of the prime ministers who have gone through different leadership periods- Deng Xiaoping, Jiang Zemin, Hu Jintao, Xi Jinping ,and summarize the focus of their governance. The research results found that the composition of the work report can be classified into eight categories: 'economic development', 'social development', 'international relations', 'government reform', 'political ideology', 'Hong Kong, Macao and Taiwan', 'national defense', 'others'. ”, the government work report during Deng Xiaoping focused on “economic development” and “international relations,” while Jiang Zemin, Hu Jintao, and Xi Jinping focused on “economic development” and “social development”; the similarities index between the years of the government work report has also maintained an average value of about 70% as the fluctuation of the international situation has slowed down, indicating that China has entered the stage of a stable developing country. Changes in the topic themes’ ratio such as 'political ideology', 'national defense', can also find differences in the power levels of different leaders, which shows the effectiveness of text mining as a research method.