檔案是國家發展的重要證據，媒體在處理重大政治議題時，常以檔案為論述依據，特別在選舉期間，媒體透過檔案相關新聞詮釋歷史，影響閱聽人對歷史事件的集體記憶及國族認同。本研究運用文本探勘技術，分析2000年至2020年間臺灣6次總統大選中，《聯合報》、《自由時報》與《中國時報》共534篇檔案相關新聞。研究採用Jieba、CKIP Tagger及LDA模型，分析報導頻率、關鍵詞詞頻及TF-IDF值，並進行主題聚類。結果顯示，選舉期間的檔案新聞多圍繞「檔案」、「歷史」、「政府」、「總統」等詞彙，並持續涉及蔣中正、蔣經國及李登輝等人物。此外，主題聚類結果表明，新聞報導框架主要集中於1999至2000年首次政黨輪替，並圍繞「一個中國」的辯論。

Archives serve as critical evidence in documenting national development and are frequently employed by the media as a foundation for political discourse, particularly when addressing significant political issues. During election periods, the media often utilize archival-related news to reinterpret historical events, influencing public collective memory and shaping perceptions of national identity. This study applies text mining techniques to examine 534 news articles related to archival issues published by three major Taiwanese newspapers—United Daily News, Liberty Times, and China Times—during six presidential elections between 2000 and 2020. By employing Jieba, CKIP Tagger, and LDA models, the research analyzes reporting frequency, keyword frequency, and TF-IDF values, while also conducting topic clustering on the news content. The findings reveal that during election periods, archival-related news frequently highlights key terms such as “archive,” “history,” “government,” and “president,” with central figures consistently featured, including Chiang Kai-shek, Chiang Ching-kuo, and Lee Teng-hui. Furthermore, the topic clustering results indicate that the reporting framework of these archival news pieces primarily focuses on Taiwan’s first political party transition between 1999 and 2000 and is closely tied to debates concerning the “One China” policy.

檔案主題新聞; 總統大選; 國族認同; 集體記憶; 文本探勘; 主題模型

Archival-themed reporting, Presidential election, National identity, Collective memory, Text mining, Topic modeling