Sentence Structure and Word Relationship Modeling for Emphasis Selection

TL;DR

This paper introduces a novel emphasis selection framework that leverages sentence structure and word relationships using graph neural networks, outperforming traditional sequence-based methods.

Contribution

It proposes a new graph-based approach incorporating sentence parse trees and word similarity graphs for emphasis selection, which is a novel direction in this task.

Findings

Achieves superior performance over traditional methods.

Effectively models sentence structure and word relationships.

Demonstrates the benefit of graph neural networks in emphasis selection.

Abstract

Emphasis Selection is a newly proposed task which focuses on choosing words for emphasis in short sentences. Traditional methods only consider the sequence information of a sentence while ignoring the rich sentence structure and word relationship information. In this paper, we propose a new framework that considers sentence structure via a sentence structure graph and word relationship via a word similarity graph. The sentence structure graph is derived from the parse tree of a sentence. The word similarity graph allows nodes to share information with their neighbors since we argue that in emphasis selection, similar words are more likely to be emphasized together. Graph neural networks are employed to learn the representation of each node of these two graphs. Experimental results demonstrate that our framework can achieve superior performance.