Minimally Supervised Learning of Affective Events Using Discourse Relations
This addresses the challenge of predicting event polarity in NLP applications, offering a minimally supervised approach that is incremental in leveraging existing discourse relations.
The paper tackled the problem of recognizing affective events in text by propagating affective polarity using discourse relations, achieving effective learning without manually labeled data and improving supervised learning with small labeled datasets.
Recognizing affective events that trigger positive or negative sentiment has a wide range of natural language processing applications but remains a challenging problem mainly because the polarity of an event is not necessarily predictable from its constituent words. In this paper, we propose to propagate affective polarity using discourse relations. Our method is simple and only requires a very small seed lexicon and a large raw corpus. Our experiments using Japanese data show that our method learns affective events effectively without manually labeled data. It also improves supervised learning results when labeled data are small.