텍스트 마이닝을 통한 향가 연구사 검토와 분석

Abstract

Recently, text mining and semantic network analysis have been used in various academic fields to produce various results. This study attempted to review the history of research by statistical methods utilizing datafication of the Hyangga papers published since 2000. This study also examined the trend and the most popular keywords of Hyangga papers in the recent two decades by extracting their titles, publication years, and keywords. Through this examination, I analyzed the strength and characteristics of the keywords and connections most frequently mentioned in Hyangga research. The most popular keywords in Hyangga research is 『Samgukyusa』. In other words, Hyangga research was used as the basis on which Hyangga was created forming a close network with 『Samgukyusa』. In addition, it was confirmed that the scope of Hyangga research is gradually expanding and deepening from Hyangga research of 『Samgukyusa』 to Hyangga research of 『Kyunyeocheon』. Network analysis, as semantic network analysis, focused on the connection of words that make up nodes. The extracted primary keywords were visualized as connected and spatial structures according to their importance. The overall meaning of Hyangga research was found by analyzing the strength and characteristics of nodes and their connections. It was also confirmed that Hyangga research was largely divided into 『Samgukyusa』, interpretation of Hyangga, Buddhism as ideological background of Hyangga, and Kyunyeo Hyangga. In addition, parody was found to be a major node related to Hyangga education as one of the recent trends of research. Furthermore, as result of analyzing the frequency and centrality of co-occurring keywords, Hyangga's work <Cheoyongga> showed the highest centrality. On the other hand, connection degree centrality in <Tosolga> and <Wonwangsangga> was higher than in <Changiparangga> and <Heonhwaga> on which independent studies were concentrated. The purpose of this study is to analyze the trends of Hyangga research by reviewing the history of Hyangga research through text mining methods that have recently developed. It is also meaningful in that this study attempts to present a new methodology and to spatially visualize the keywords of research papers to derive the trends of Hyangga research. The data mining analyzing 100 years of Hyangga research remains a challenge for further study.