Neural Transition-based Syntactic Linearization

TL;DR

This paper introduces a neural transition-based syntactic linearizer that outperforms LSTM language models in generating grammatically correct sentence orders with syntactic structure.

Contribution

It presents a novel neural transition-based approach using a feed-forward network for syntactic linearization, achieving superior results over existing LSTM models.

Findings

Neural transition-based linearizer outperforms LSTM language models.

Feed-forward neural network effectively models syntactic linearization.

Significant improvement in grammatical sentence ordering.

Abstract

The task of linearization is to find a grammatical order given a set of words. Traditional models use statistical methods. Syntactic linearization systems, which generate a sentence along with its syntactic tree, have shown state-of-the-art performance. Recent work shows that a multi-layer LSTM language model outperforms competitive statistical syntactic linearization systems without using syntax. In this paper, we study neural syntactic linearization, building a transition-based syntactic linearizer leveraging a feed-forward neural network, observing significantly better results compared to LSTM language models on this task.