誰替記憶上色？維基百科上的集體記憶形成動態

Abstract

在數位時代下，維基百科（Wikipedia）已成為全球最具影響力的數位知識平台，其為當今社會中重要的集體記憶載體。然而，過往研究多聚焦於單一面向，較少同時整合內容表徵、編輯歷程與公眾關注度等。本研究旨在填補此研究缺口，將維基百科視為重要記憶平台，選取中世紀至今的五個重大歷史事件（十字軍東征、第一次世界大戰、第二次世界大戰、九一一事件以及新冠疫情）作為研究對象，系統性比較20種語言版本的維基百科條目。本研究透過MediaWiki API擷取固定時間的內容條目、條目創立至今的完整編輯歷程以及2015-2026年的每日頁面瀏覽量，研究結果呈現三大核心發現，首先，在記憶原料層面，久遠事件高度依賴書籍著作，當代事件則依賴即時新聞，且跨語言的來源與流向揭示資訊高度不平等；第二，在編輯權力層面，數據呈現高Gini係數與前5%編輯者貢獻占比，證實知識生產受控於少數核心群體，透露出集體記憶並不集體的現象；最後，在注意動態層面，證實公眾對歷史記憶的喚起有選擇性，且不同語言社群有不同的注意力波動狀態。綜上所述，本研究有助於重新思考當今時代下的集體記憶現象，並為資訊不平等與知識權力分配提供實證啟示。

In the digital age, Wikipedia has become the world's most influential digital knowledge platforms, serving as an important carrier of collective memory in contemporary society. However, past research has largely focused on a single aspect, rarely integrating content representation, historical editing, and public attention simultaneously. This study aims to fill this research gap, treating Wikipedia as a crucial memory platform and selecting five major historical events from the Middle Ages to the present (the Crusades, World War I, World War II, 9/11, and the COVID-19 pandemic) as research subjects, systematically comparing Wikipedia entries in 20 language versions. This study utilizes MediaWiki. The API extracted content entries over a fixed period, their complete editorial history from creation to the present, and daily pageviews from 2015 to 2026. The research revealed three core findings: First, at the level of memory raw materials, long-standing events heavily rely on books, while contemporary events rely on breaking news, and cross-language sources and flows reveal a high degree of information inequality; second, regarding editorial power. At the data level, the high Gini coefficient and the proportion of the top 5% editors share confirm that knowledge production is controlled by a small core group, revealing that collective memory is not truly collective. Finally, at the level of attention dynamics, it is confirmed that the public has a selective ability to recall historical memories, and that different language communities exhibit different attentional fluctuations. In conclusion, this study helps to rethink the phenomenon of collective memory in the present era and provides empirical insights into information inequality and the distribution of knowledge power.