Research on creation of language resources for highprecision modality analysis in Japanese

点击翻译按钮获取中文摘要

I have constructed language resources and tools for Japanese modality analysis. They include the following three main components: (1) a corpus of 41,799 eventswith extended modality tags, which are extracted from the latest version of “Balanced Corpusof Contemporary Written Japanese”, (2) a corpus of 1,785 pairs of negation triggers and their foci, and, (3) a detector of the focus of a negation trigger in Japanese, which achieved accuracy of 80 % in the domain of posted reviews about hotel service.