場合：The Pacific Neighborhood Consortium 2015 Annual Conference and Joint Meetings
投影片：slides
Tools for textual analysis provide opportunities for researchers to study large corpora from a wide variety of perspectives, thereby opening windows to observe new findings that were hard to achieve before. In this talk we overview some of such applications. Machine learning methods helped us identify biographical information in literary Chinese in Difangzhi (地方志), which can be useful for enlarging the contents of the China Biographical Database hosted by Harvard University. Analysis of word collocation and topic modeling helped us analyze various types of human-rights issues in the Renmin Ribao of China. Word frequencies proved instrumental for showing the development of the 228 incident in Taiwan. Contextual analysis of a Hakka word “硬頸” accounted the stories about how semantics of words could change over time and how the changes spread to different genres of media in Taiwan. Similar analyses shed light on how Chinese people referred to “China” using words like “天下” and “中國” in La Jeunesse(新青年). Computational analysis of the Complete Tang Poems (全唐詩) led us to find that “white” is the most frequent color in the poems, and allowed us to compared the styles of many poets.