2009 Fiscal Year Final Research Report

Detecting Conclusion-Evidence Relations in Web Texts Using Anaphora Resolution Technique

Research Project

Project/Area Number	20700138
Research Category	Grant-in-Aid for Young Scientists (B)
Allocation Type	Single-year Grants
Research Field	Intelligent informatics
Research Institution	Tokyo Institute of Technology
Principal Investigator	IIDA Ryu Tokyo Institute of Technology, 大学院・情報理工学研究科, 助教 (40464276)
Project Period (FY)	2008 – 2009
Keywords	照応解析 / 根拠帰結関係同定
Research Abstract	This research focuses on the task of detecting Conclusion-Evidence relations in texts, which consist of evaluative expressions such as "good" and "excellent" and their evidence expressions. We have developed a technique of resolving anaphoric relations in texts, which is one of clues to capture discourse relations occurring in texts. In order to incorporate such anaphora resolution technique into Conclusion-Evidence relation detection, we refer to the ideas exploited in the works by Iida et al. (2004) and Iida et al. (2005). As we conduct empirical evaluation of Conclusion-Evidence relation detection, we reported that our model achieved better performance of the relation detections appearing in both Web texts and newspaper articles in comparison to a baseline model.