Serial Number: http://www.riss.kr/link?id=A107047926

Title: Exploring The Sijo Research Method Using Digital Analysis Technique -Based On The Monotype Sijo In The “Korean Sijo Dictionary”-

Author: Kim Seong Moon, Kim Ba-ro

Journal: Journal of Multi-Cultural Contents Studies

Vol: (34)

Pages: 209-235

Date: 2020.

Register Information: KCI

etc.: -
-----------------------------------------------------------------------

<Abstract>

Humanity is facing a drastic change called the Fourth Industrial Revolution. This is changing human life, and literary research is no exception. In line with these changes, this paper is an attempt to explore new classical research methodologies for the changing times in terms of the use of digital analysis techniques. For this purpose, out of 4,736 sijo works contained in the “Korean Sijo Dictionary” (Park Eul-Su, Asiaculturehistory publishing house, 1992), 4,127 sijo works were conducted, with the exception of janghyeong sijo, to conduct digital analysis using hierarchical analysis and deep learning.
First of all, through hierarchical analysis, 13 writers with more than 30 episodes were analyzed for the similarity of the progenitor, and as a result, four groups with similar vocabulary and vocabulary combinations were extracted. Next, through deep learning, we have implemented a model that can predict the age of creation of sijo for works from 1600 to 1800s. Although the accuracy of the predictions was not the same for each period, it was possible to predict the date in which the unknown works of the artist were created in the future and to examine the characteristics of the works by regiment. Finally, we took a step further from the traditional search method in which only the morphemes were able to search for matching sijos, and implemented deep learning semantic search for sijo. This enabled the extraction of works with similar meaning in content in order of high similarity, even if the keywords or sentences that were to be searched and the morphemes did not match. I think this will be meaningful in that it will significantly reduce the time required to classify and analyze vast amounts of sijo works by subject or keyword.
Although research methods using digital analysis techniques discussed above may not replace traditional research methods, it is hoped that they will be able to secure their own significance in that they are seeking and attempting new research methods in line with the trend of the new era.