人物與機構之知識圖譜分析工具發展與數位人文應用

Abstract

知識圖譜為語意上資料結構的一種，已被證實有助於數位資訊系統進行更有意義的知識表達，並且憑藉其動態鏈結與視覺化的特性，更有利於知識的引導與吸收。本研究旨在開發支援數位人文研究之「人物與機構之知識圖譜分析工具」，輔以人文學者透過知識圖譜功能的特性，不僅清楚看到人物與人物之間的關聯，更能透過人物與機構關係看到潛在於文本中的人物關係。為了驗證此一工具對於支援數位人文研究的效益，本研究以準實驗研究法之對抗平衡設計比較實驗對象依序使用有無「知識圖譜功能之人物與機構關係分析工具」進行文本人物與機構脈絡探索，在探索成效與效率上是否具有顯著的差異；並以科技接受度問卷、半結構訪談的方式瞭解實驗對象對「人物與機構之知識圖譜分析工具」的看法與感受；最後，透過滯後序列分析搭配螢幕錄影，探討實驗對象操作兩系統時的行為轉移差異。\n\n實驗結果發現，相較於採用「無知識圖譜功能之人物與機構關係分析工具」，採用「具知識圖譜功能之人物與機構關係分析工具」更能輔助受測對象在有限的時間內探索文本中的人物與機構關係，提高文本脈絡探索成效。在科技接受度分析得知，受測對象對於「具知識圖譜功能之人物與機構關係分析工具」持正面肯定態度，認為此工具的操作直觀且能幫助到他們進行文本人物與機構脈絡探索。此外，滯後序列分析結果發現，使用「具知識圖譜功能之人物與機構關係分析工具」的受測對象能在短時間內找到有系統性的探索模式，進而完成人物與機構脈絡探索表所賦予的任務。訪談內容分析得知大部分受測對象皆認為知識圖譜工具能做為探索人物與機構關係時很好的探索切入點，提高整體人物脈絡探索的成效與效率，但也有部分受測對象認為知識圖譜工具在一次觀看多位人物時，其圖上呈現的資訊量有點過多，認為知識圖譜呈現功能還有進一步改善空間。在未來研究方向上，可以考慮納入更多實體元素，舉凡地名、時間等充實知識圖譜，並且引入更多南洋人物誌或名人傳記文本充實平台內容。

Knowledge graph is a kind of semantic data structure, which has been proved its benefits in promoting digital information system to carry out more meaningful knowledge representation, and by virtue of its dynamic link and visual characteristics, it is more conducive to knowledge guidance and absorption. This research aims to develop a Knowledge Graph Analysis Tool of Characters and Institutions (KGAT-CI) that can support digital humanities research more effectively. Knowledge graph in the KGAT-CI is provided for humanities scholars so that they not only clearly see the relationship between characters, but also view the potential character relationships through characters and institutional connection. In order to verify the effectiveness of this tool in supporting digital humanities research, a counterbalanced design in the quasi-experimental research was applied in this study to compare the experiment subjects of two groups who respectively used a character digital humanities research platform with and without KGAT-CI to explore the relationship between the two entities of character and institution, and if there were significant differences in the learning effectiveness and efficiency between the two groups. Technology acceptance questionnaire and semi-structured interview were utilized for understanding the experiment subjects’ opinions and perception toward KGAT-CI. Finally, lag sequential analysis and screen recording analysis were used for observing the experiment subjects’ behavior processes using two different systems to discuss whether the notable difference in the operation behavior transfer between the two groups existed.\n\nThe experimental results show that the KGAT-CI could help the experiment subjects to improve the effectiveness of exploring the relationships between characters and institutions under limited time. The technology acceptance of the experiment subjects with KGAT-CI support reveals highly positive satisfaction. It presents that such a tool could help the experiment subjects explore the connections between characters and institutions. Besides, lag sequential analysis reveals that the experiment subjects who used KGAT-CI could rapidly generalize systematic patterns to explore the relationship between characters and institutions from texts in limited time. The interview reveals that most of the experimental subjects believed the knowledge graph can be a good entry point for exploring the relationship between characters and institutions, improve the effectiveness and efficiency of the overall character network exploration. However, some experimental subjects expressed that the KGAT-CI provides too much information when this tool was used to view multiple characters and institutions at a time. In the future research directions, it is able to including more physical elements, such as location, time, etc. to enrich the knowledge graph in the KGAT-CI, and import more Southeast Asia characters or celebrity biographies to enrich the character digital humanities research platform.