文本探勘用之於日語相關人文類研究方面的質性研究應用與課題

Qualitative Research in the Humanities on Japanese Language: Applications and challenges of text mining methods

目前日語相關研究上使用的計量語言學研究方法，大多偏向計量語言學領域與課題。而應用於其他日本語學、文學相關研究領域上的資料探勘（文本探勘），卻仍然非常有限。由於當今語料庫或電子資料的普及，逐漸形成將之廣泛應用於各領域研究的趨勢。本論文乃是利用第三代AI技術之自然言語處理重要領域的資料探勘手法，特別是善用從龐雜的資料中擷取有意義的資訊之文本探勘技術，嘗試探討應用於日語以及日語教育等日語相關人文類研究的質語。研究結果顯示：以文本探勘技術的量化分析為依據，反而可以將蘊藏於文本中的多元化以及多層次的意義（建立文脈組織的質化意味的集合）等質性分析結果彰顯出來。並可以確定透過導入文本探勘技術，將多元性且客觀性的成果呈現出來。

Up to the present in research related to the Japanese language, the quantitative linguistic method is limited to quantitative linguistic disciplines and issues. In other areas of Japanese language studies, the application of data mining (text mining) techniques has been quite limited. However, with the widespread use of corpora and electronic data, there has been a movement towards research using them. In this paper, I focus on text mining techniques to extract meaningful material by applying data mining methods to linguistic materials, which is a field of natural language processing that is an important part of the third AI technology. This study is an attempt at text mining into a qualitative study of Japanese language related humanities studies, including Japanese language studies and Japanese language education. As a result, when performing qualitative analysis, it is possible to trace the layers of diverse and layered semantic spaces (sets of qualitative semantic relations that make up a context) contained in a text by using text mining as a cue. The application of text mining into humanities research in the Japanese language paves the way for more diverse and objective results.