2021-01-012024-05-13https://scholars.lib.ntu.edu.tw/handle/123456789/654126"經由系統性的讀取文本並將拆解之片段內容進行歸類，自動化文本分析突破傳統質化內容分析的限制，使研究者得以從文本中推測文字潛在的意涵。近年來運算性文本分析研究法的進展替有意分析涵蓋多個潛在論述主題的高維度文本資料的研究者提供了嶄新的研究途徑。最新的相關研究更實現了透過文本資料的分析將政治人物的政治立場投射於意識形態的光譜上。 受到此一資料驅動的嶄新研究途徑的啟發，本子計畫提出了一個整合性的途徑透過臉書發言資料的爬取去分析2020總統大選中候選人的政策立場。本研究的長處在於使用了一系列運算性文本分析研究工具去發掘政黨/候選人在選戰過程中的議題爭議面向，估計此些政治行為者相對的意識形態立場，並透過線上問卷的執行了解候選人與其社群媒體追隨者在不同議題立場上的立場分歧與聚合，以及這背後的社經因素。 計畫的目的不僅止於使用新式資科研究方式去回答國內政治過程中產生的傳統理論議題。我們的最終目的在於形塑一個科際整合的研究途徑去檢視社群媒體對當前政治的影響；經由模型的建構、假說驗證、並使用本國特有的科技所發展的研究工具驗證理論問題，使產出的研究成果能供廣大的社會科學研究者做檢證。" "Through systematic reading and assigning pieces of content to pre-coded conceptual categories, automated content analysis allows researchers to infer latent meaning from text that would not otherwise possible in traditional qualitative content analysis. Recent advance in computational content analysis has provided researchers promising new ways to analyze high-dimensional text data that contain multiple topics of discourse. More recent studies even demonstrate the plausibility of mapping the ideological positions of political actors on a multi-language platform. Inspired by this emerging scholarship of data-driven research, this subproject proposes an integrated approach toward analyzing political candidates’ policy positions on a multi-dimensional campaign issues space based on messages crawled from their Facebook accounts in the upcoming 2020 Taiwan Presidential election cycle. The strength of this study lies in using a variety of computational content analysis tools to unpack the dimensionality of conflict throughout the campaign, estimate candidates’ relative positions across these issue dimensions, and assess the link between candidates’ observed issue convergence/divergence and their social media followers’ profiles via online-administered surveys to better comprehend the socio-economic foundation of policy positioning in an age of social media. The purpose of this subproject goes well beyond applying new methods to answer theoretical puzzles unique to domestic political process. Our ultimate goal is to forge an interdisciplinary collaboration toward examining the political impact of social media by iterating through stages of model development, hypothesis testing, and validation with emphasis placed on using tools developed by Taiwan’s own technology and generate replicable research results that can be evaluated by the wider social science community internationally."社群媒體資料科學機器學習文本分析問卷選舉Social mediadata sciencemachine learningtext analysissurveyelection