From Textual to Historical Networks: Social Relations in the Biographical Dictionary of Republican China

In this paper, we combine natural language processing (NLP) techniques and network analysis to do a systematic mapping of the individuals mentioned in the Biographical Dictionary of Republican China, in order to make its underlying structure explicit. We depart from previous studies in the distincti...

Description complète

Enregistré dans:  
Détails bibliographiques
Auteurs: Armand, Cécile (Auteur) ; Henriot, Christian 1955- (Auteur)
Type de support: Électronique Article
Langue:Anglais
Vérifier la disponibilité: HBZ Gateway
Journals Online & Print:
En cours de chargement...
Fernleihe:Fernleihe für die Fachinformationsdienste
Publié: Université du Luxembourg 2021
Dans: Journal of historical network research
Année: 2021, Volume: 5, Numéro: 1, Pages: 114-153
Sujets non-standardisés:B Elites
B NLP
B China
B Biographical dictionary
B Co-occurrence
Accès en ligne: Volltext (kostenfrei)
Volltext (kostenfrei)
Description
Résumé:In this paper, we combine natural language processing (NLP) techniques and network analysis to do a systematic mapping of the individuals mentioned in the Biographical Dictionary of Republican China, in order to make its underlying structure explicit. We depart from previous studies in the distinction we make between the subject of a biography (bionode) and the individuals mentioned in a biography (object-node). We examine whether the bionodes form sociocentric networks based on shared attributes (provincial origin, education, etc.). Our major contribution consists in annotating the links between individuals in order to (1) question the assumption that word cooccurrences equate with actual relations; (2) define a more accurate classification of relationships among elites in republican China. We demonstrate that political and professional relations in this population outweigh the types of social ties commonly accepted in the scholarship on modern China. We eventually develop a method that can be applied to similar corpora in a critical and comparative perspective.
ISSN:2535-8863
Contient:Enthalten in: Journal of historical network research
Persistent identifiers:DOI: 10.25517/jhnr.v5i1.117