A Bootstrap Method for Automatic Rule Acquisition on Emotion Cause Extraction
Shuntaro Yada, Kazushi Ikeda, Keiichiro Hoashi, Kyo Kageura · 2017
Emotion cause extraction is one of the promising research topics in sentiment analysis, but has not been well-investigated so far. This task enables us to obtain useful information for sentiment classification and possibly to gain further insights about human emotion as well. This paper proposes a bootstrapping technique to automatically acquire conjunctive phrases as textual cue patterns for emotion cause extraction. The proposed method first gathers emotion causes via manually given cue phrases. It then acquires new conjunctive phrases from emotion phrases that contain similar emotion causes to previously gathered ones. In existing studies, the cost for creating comprehensive cue phrase rules for building emotion cause corpora was high because of their dependencies both on languages and on textual natures. The contribution of our method is its ability to automatically create the corpora from just a few cue phrases as seeds. Our method can expand cue phrases at low cost and acquire a large number of emotion causes of the promising quality compared to human annotations.