Japanese Predicate Argument Structure Analysis Exploiting Argument Position and Type

Abstract

We propose an approach to Japanese predicate argument structure analysis exploiting argument position and type. In particular, we propose the following two methods. First, in order to use information in the sentences in preceding context of the predicate more effectively, we propose an improved similarity measure between argument positions which is more robust than a previous co-reference-based measure. Second, we propose a flexible selection-and-classification approach which accounts for the minor types of arguments. Experimental results show that our proposed method achieves state-ofthe-art accuracy for Japanese predicate argument structure analysis.