Developing system for supporting structuration of public opinions based on visualization of topic transition and rhetorical structure theory

点击翻译按钮获取中文摘要

We built an argumentation corpus consisting of 506 speakers and 1,764 utterances. The corpus was automatically converted from Wikipedia discussion note pages to RDF (Resource Description Framework) and manually annotated with rhetorical relations. The corpus was used for analyzing appropriateness of argumentation structure. This analysis showed that appropriate argumentation structures have higher probability of being followed by agreement utterance. Moreover, we generalize an actual instance of concern assessment in Aioiyama committee and generate a prototype of ontology for concern assessment.