Greedy Transition-Based Dependency Parsing with Discrete and Continuous Supertag Features

TL;DR

This paper enhances greedy transition-based dependency parsing by integrating both discrete supertag features and continuous vector representations, leading to improved accuracy on the English Penn Treebank.

Contribution

It introduces a novel approach combining discrete and continuous supertag features, achieving state-of-the-art results in greedy dependency parsing.

Findings

Achieved 88.6% LAS and 90.9% UAS on English Penn Treebank.

Adding continuous supertag representations improves parsing accuracy.

Demonstrated the effectiveness of combined feature types in dependency parsing.

Abstract

We study the effect of rich supertag features in greedy transition-based dependency parsing. While previous studies have shown that sparse boolean features representing the 1-best supertag of a word can improve parsing accuracy, we show that we can get further improvements by adding a continuous vector representation of the entire supertag distribution for a word. In this way, we achieve the best results for greedy transition-based parsing with supertag features with $88.6\%$ LAS and $90.9\%$ UASon the English Penn Treebank converted to Stanford Dependencies.