This paper applies plot analysis, a traditional method of humanities narrative analysis, to realize a quantitative narrative structure and conducts a quantitative analysis of the result. The validities of plot divisions and a classification are evaluated numerically by calculating identity frequencies through plot analysis which is a technique employed within the humanities. The serial patterns of the narrative structures are extracted by utilizing the n-gram method on the results of a plot classification. Period transitions within frequent plots are also extracted using the chi-square test. Moreover, theme words for the stories are extracted using a quantitative technique that analyzes the relation between theme and plot, and the stories are subsequently classified according to the themes. The plot characteristics for each theme are also extracted based on the theme-classification results. This is not a completely automatic method because the divisions and the classification were created manually, albeit by utilizing the quantitative index. However, it is expected that the techniques reported in this paper can form the basis of a completely automatic method for analyzing stories in the future.

View full abstract

The aim of the present paper is to scientifically verify the stylistic changes of one author of novels. The target data consist of the twelve full-length works of Haruki MURAKAMI, as a representative of modern Japanese literature. We measured vocabulary distributions for the texts, extracted feature vectors from them using classifications in terms of both word-class and semantic categories, and conducted clustering analyses on the data in order to identify groupings based on style within his twelve full-length works. The analysis based on word-class yielded clusters that reflect a diachronic division, indicating that MURAKAMI's style has shifted over time. In contrast, the analysis based on semantic categories yielded clusters that reflect the themes and the contents of the novels, indicating that MURAKAMI's interest has shifted from the individual to society.

View full abstract

Recently in Japan, many computer-assisted systems to motivate people to continue walking exercise have been used for health promotion. Some of them try to motivate people using the mechanisms of SNS; they expect that the comradery and the communications among users in SNS systems motivate users to continue their walking exercise. Those systems, however, have a limitation that only the users who afford to make text-based conversation can use the systems. Thus in this paper, we designed a SNS based system to support continuous walking exercise using the number of daily walking steps instead of text-based conversation. Our system is designed in consideration of three motivation levels with three motivation mechanisms to provide various motivation effects without enforcing users so much labor. Through our experiment it is shown that the computer-assisted systems can provide proper motivation effects to have users continue their walking exercise.

View full abstract