A study was conducted to examine the appropriateness of text mining of martial arts poems to extract their semantic content relating to the philosophy of martial arts in the context of iaijutsu. The target materials were 180 poems from 7 iaijutsu schools. Using KH Coder analysis software, frequently occurring words were extracted, and hierarchical cluster analysis was performed. The results obtained were as follows.
　(1) The most frequently occurring nouns were “heart” (48 times), “iai” (25 times), and “enemy” (24 times), whereas frequently occurring verbs were “pull out” (32 times), “cut” (31 times), and “know” (29 times).
　(2) Hierarchical cluster analysis generated 13 clusters. The philosophy of iaijutsu, as interpreted by 13 typical poems, is: “The purpose of iaijutsu is only to cut, to cut down at once, and make the opponent become a buddha. The key is to overcome oneself.” The art of iaijutsu was interpreted as “Always confront the opponent with an ordinary mind (no thought) and with the mind's eye. Instead of drawing the sword, one should promptly push the sword out to the opponent's shoulder and cut him down in a single stroke. It is also important to pass the sword off at times.”
　(3) The typical poems in each cluster were classified according to the 11 themes indicated by K. Kobayashi (2004): “(iai) philosophy,” “physical technique”, “mental technique”, “budo spirit,” and “training spirit”. Comparing these results with those of K. Kobayashi (2004), I found that nearly 90% of the themes in the martial arts poems agreed.
　Based on the above results, I concluded that application of text mining is a reasonable and efficient approach for grasping the semantic content of martial arts poems. However, since this study was conducted on 180 poems relating a specific martial art, it will be necessary to study more such poems and other martial arts to confirm the validity of this method and further improve its accuracy and efficiency.