Distance-Free Modeling of Multi-Predicate Interactions in End-to-End Japanese Predicate-Argument Structure Analysis

TL;DR

This paper introduces novel Japanese PAS analysis models that utilize pooling and attention mechanisms to directly capture interactions among multiple predicate-argument structures without relying on word order or distance, improving accuracy.

Contribution

The paper proposes distance-free models for Japanese PAS analysis that extend bi-RNNs with pooling and attention to better capture multi-PAS interactions.

Findings

Improved prediction accuracy for indirect dependency cases

Achieved new state-of-the-art F1 score on benchmark corpus

Models effectively capture multi-PAS interactions without word order dependence

Abstract

Capturing interactions among multiple predicate-argument structures (PASs) is a crucial issue in the task of analyzing PAS in Japanese. In this paper, we propose new Japanese PAS analysis models that integrate the label prediction information of arguments in multiple PASs by extending the input and last layers of a standard deep bidirectional recurrent neural network (bi-RNN) model. In these models, using the mechanisms of pooling and attention, we aim to directly capture the potential interactions among multiple PASs, without being disturbed by the word order and distance. Our experiments show that the proposed models improve the prediction accuracy specifically for cases where the predicate and argument are in an indirect dependency relation and achieve a new state of the art in the overall $F_1$ on a standard benchmark corpus.