Comparative Analyses of Textual Contents and Styles of Five Major Japanese Newspapers

摘要

Newspapers remain an important media from which people obtain a wide variety of information. In Japan, there are five major newspapers, having their own opinions and ideologies. Although these are readily recognized, they are infrequently investigated from the viewpoint of their textual characteristics. This study analyzes these differences among the five newspaper editorials. We apply morphological analysis and count the frequency of morphemes within the text data. We then apply principal component analysis and random forests classification experiments to examine their similarities and differences. Throughout these statistical analyses, we use function words and content words as features, which enables us to determine which of the two characteristics -styles or content- more powerfully affects the classification types. This study contributes to text classification studies by deliberately comparing the classification performances provided by different feature sets, function words and content words. In addition, this study will provide an empirical basis for understanding the similarities and differences among the five newspapers.