作者: Kazi Saidul Hasan , Vincent Ng
DOI: 10.3115/V1/D14-1083
关键词:
摘要: Recent years have seen a surge of interest in stance classification online debates. Oftentimes, however, it is important to determine not only the expressed by an author her debate posts, but also reasons behind supporting or opposing issue under debate. We therefore examine new task reason this paper. Given close interplay between and classification, we design computational models for examining how automatically computed information can be profitably exploited classification. Experiments on our reason-annotated corpus ideological posts from four domains demonstrate that sophisticated stances indeed yield more accurate results than their simpler counterparts.