AI人機協作應用於系統文獻探討之研究──以漫畫史為例（2000~2025）

AI-Human Collaboration in Systematic Literature Reviews: A Case Study of Comic History (2000–2025)

本研究旨在探討人工智慧（Artificial Intelligence，AI）在人機協作下應用於系統性文獻回顧（Systematic Literature Review，SLR）的可行性與方法論貢獻，並以2000年至2025年間華文學術期刊中的「漫畫史」研究為案例。研究首先依循PRISMA 2020與PRISMAS規範，建立嚴謹之檢索與篩選流程，最終納入62篇具代表性的期刊論文。其次，透過大型語言模型（Large Language Models，LLM）進行語義單位萃取與初始編碼生成，再由研究者人工審核以確保準確性與一致性。研究結果顯示，漫畫史研究的八大主軸呈現多核心並進之格局，其中「史觀與制度政策」、「文本美學與歷史敘事」以及「通路與經營」為最具增長力的主導軸。AI在語義前處理與數據統整上顯著提升效率與透明度，但在語境詮釋與事件波動分析上，仍須仰賴研究者判斷。本文不僅提供一套「人審—機審—再人審」的操作模板，也展示AI如何在提升研究可追溯性與效率的同時，維持質性分析的反思性與深度。

This study examines the role of Artificial Intelligence (AI) in supporting Systematic Literature Reviews (SLR) through human–machine collaboration, using Chinese-language“comic history”research (2000–2025) as a case study. Following PRISMA 2020 and PRISMAS guidelines, 62 journal articles were identified through a rigorous search and screening process. Large Language Models (LLMs) assisted in extracting semantic units and generating initial codes, which were verified through manual auditing for accuracy and consistency. The results reveal a multi-core landscape of comic history scholarship, with notable growth in historical perspectives, textual aesthetics, and institutional-industry analyses. AI improved efficiency and transparency in preprocessing and data structuring, while contextual interpretation and event-driven variations still required human judgment. The study proposes a“human review–AI processing–human verification”workflow that enhances traceability and efficiency while retaining the reflexivity and depth of qualitative analysis.