Historical inference based on semi-supervised learning

摘要: Abstract In the past, most historical research has been manually carried out by exploring facts reading between lines of documents. Nowadays, big data become electronically available and advances in machine learning techniques allow us to analyze vast amount data. From a perspective, making inferences about political stances figures is important for grasping rivalries power structures an era. Thus, this paper, we propose approach systematic inference mechanisms based on human network constructed from network, humans are linked according degree kinship using genealogy records, identified agendas recorded annals dynasty as force. And then, algorithm, semi-supervised learning, classifies who cannot identify forces that reflect links networks. The consist Andong Gwon clan, record family relations 10,243 people 10th 15th century Korea, Annals Joseon Dynasty, volume describes Dynasty 472 years composed 1894 fascicles 888 books. data, construct historically meaningful period (1443–1488), classify into two proposed method. We suggest study could be utilized potent reference tool devoid subjectivism experts field history.