Return to search

由史料中探勘社會網絡:以乾隆時期為例 / Social Network Mining from Historical Documents- by Example during Qianlong's Reign

古今中外歷史中,在政治權力的結構裡,除了在最上位的領袖外,其下的文武百官,往往根據其職份或私交等情況,自成人際關係網絡。然而,依照每個人在網絡中位置的不同,重要程度也有所不同。在網絡中扮演重要角色者,除了代表其人際關係愈複雜外,同時也暗示其政治影響力愈大。這些人物往往也就是足以影響當時政治的「權臣」。然而在歷史上,有些皇帝的在位時間較長、統治時間較久。在其統治期間,可能因為皇帝本身,或政治環境遞嬗等因素,使得不同的時期有不同「權臣」,或是其晚年才出現明顯的「權臣」。本論文便是基於這樣的歷史現象,研究由史料中探勘當時的人脈網絡。我們先從文本中,自動擷取出人名。然後,藉由人物在文本中與其他人物的共現場合,建立歷史人物的人脈網絡,接著利用社會網絡分析的理論基礎,分析這些網絡,進而在網絡中找出權臣,以及偵測政治權力結構的變化,為時代作出分期。本研究選用的文本為《清實錄》中的《高宗純皇帝實錄》,意欲以清高宗 (乾隆)時期為例,探勘該朝的人脈網絡,完成上述之工作。希望這樣的研究,可以在中國政治制度史等研究上,協助史學研究者。 / In power structure from ancient times to the present, officials who under the leader usually take part in the social network according to their positions or friendship. However, the importance of each person is different by their locations in network. The people who play important roles in network have complex interpersonal relationship as well as high influence in political situation. We call them "chief counselors." But in the history, some emperors reign for extremely many years. Due to some causes, such like emperor himself or changing of political circumstances, there could be several different chief counselors during their reign. This thesis focuses on social network mining from historical documents in view of above-mentioned historical phenomenon. After extracting person names from the corpus, we can construct social network by co-occurrence of people, then to find chief counselors and detect transition of power structure by Social Network Analysis. The "Veritable Records of Gaozong" is taken as the example for experiments and the result of effectiveness analysis demonstrates that the proposed methods are helpful to assist historian for historical research.

Identiferoai:union.ndltd.org:CHENGCHI/G0957530141
Creators朱政吉, Chu, Cheng Ji
Publisher國立政治大學
Source SetsNational Chengchi University Libraries
Language中文
Detected LanguageEnglish
Typetext
RightsCopyright © nccu library on behalf of the copyright holders

Page generated in 0.0026 seconds