本研究旨在開發支援數位人文研究之「觀點變遷和年代劃分工具」系統， 輔助使用者於短時間內快速掌握具有時間連續性相關文本群中「出現於文本內容中的特定主題」脈絡，不必花費過多時間閱讀、篩選文本群中「未出現 欲探勘主題的文本」，且能找出先前未想到的關聯、未想到應該去閱讀的文本。本工具能將具有時間連續性的相關文本，切分為數個時期，並以詞向量 作為探勘輸入之關鍵字在各時期中關聯程度最高的關聯詞彙依據，使用者依 據結果搜尋關聯詞彙所出現的文本，從而瞭解這些文本與欲探勘主題間的潛 在關聯；並以分群方式協助使用者分析時期是否需要分裂或合併，使用者依 此可修正原先對文本所下的時期切分，以對整體文本可分為的時期有更全面 性的理解。透過上述兩項功能，協助使用者短時間內掌握時間連續性的相關 文本中「出現於文本內容中的特定主題」的脈絡。本研究以實驗研究法比較 研究對象（具人文、歷史、哲學、民族背景研究者11 人，其中碩士以上2 人、碩士生8 人、大學生1 人）使用此系統後對於「出現於文本內容中的特定主 題」的理解是否具有顯著提升，輔以科技接受度問卷、半結構式訪談等了解實驗對象對此系統的感受。實驗結果發現，對於「出現於文本內容中的特定主題」的掌握度具有顯著的提升。在系統滿意度調查中，除訪談資料對系統輔助分析文本的便利性予以肯定外，科技接受度量表（六點量表，其中1 為非常不同意、6 為非常同意）於認知有用性平均分數為5.17 分、認知易用性平均分數為4.39 分。由訪談資料及科技接受度量表得，實驗對象對於此系統持高度肯定的正面評價。

This study aims to develop “Observation Tool of Viewpoint Migration and Age Division” for supporting research on digital humanities, allowing humanists dividing texts with time continuity into several phases to train word vectors. Word vectors are regarded as the reference for relational vocabulary with the highest relevance in various phases to mine input keywords. Clustering is also applied to assist users in analyzing whether two neighboring phases require split or merge to determine better perspective change observation vocabulary and phase division. Meanwhile, real-time interaction interface is provided for assisting users, through human-computer interaction, in grasping a wide range of contexts with specific subjects in texts with time continuity. Experimental research is applied in this study to compare the significant promotion of comprehension of specific subjects in texts and the remarkable change in cognition of research subjects before and after using “Observation Tool of Viewpoint Migration and Age Division”. Technology acceptance questionnaire, and semi-structured interview are utilized for understanding the experiment objects’ opinions and perception of “Observation Tool of Viewpoint Migration and Age Division”. The experimental results reveal that the experiment objects, after using “Observation Tool of Viewpoint Migration and Age Division”, significantly promote the grasp of specific subjects in texts. Furthermore, the interview data and technology acceptance scale reveal highly positive evaluation of the experiment objects on “Observation Tool of Viewpoint Migration and Age Division”. It presents that such a tool could rapidly and conveniently allow them to understanding the context of specific subjects in large amount of texts with time continuity.

數位人文；文本探勘；資訊視覺化；羅家倫先生文存

Digital humanities; Text mining; Information visualization; Works of Mr. Lo Chia-lun