Abstract
This paper is a research report as a sequel to "A Quantitative analysis into Waka collection and an Approach by machine learning” presented at the 25th Annual Conference of Japan Society of Information and Knowledge in 2017. The target of analysis are thirteen Imperial-commissioned anthologies which is called "Jyu-Sandai-Shu" from the early Kamakura period to the middle Muromachi period.Since the previous research was “Hachidai-Shu”, we completed the measurement of twenty one Imperial-commissioned anthologies.
We analyzed in the same way as our previous study. We performed a quantitative analysis using morphemes and examined the difference of “Jyu-Sandai-Shu”. We investigated as a method of analysis, using statistical difference of part-of-speech usage rate among “Jyu-Sandai-Shu”, especially seasonal poems and love poems, using the relationship between noun rate and MVR . Then, we tried to discriminate by using the Naïve-Bayes Method which is one of machine learning methods.
Furthermore, since this era is the era of the “Bushi’s Government”, “Bushi’s Waka" are selected more by Imperial-commissioned anthologies.We attempted to analyze the characteristics of “Bushi’s Waka" by principal component analysis method.
