# Improving a Strong Neural Parser with Conjunction-Specific Features

## TL;DR

This paper enhances a neural dependency parser by adding conjunction-specific features, significantly improving the accuracy of attaching conjunct relations, especially in coordination structures, on standard benchmarks.

## Contribution

It introduces conjunction-specific features into a neural parser, focusing on conjunct head word similarity, to improve coordination attachment accuracy.

## Key findings

- Improved 'conj' attachment accuracy
- Enhanced overall dependency parsing performance
- Effective use of conjunct similarity features

## Abstract

While dependency parsers reach very high overall accuracy, some dependency relations are much harder than others. In particular, dependency parsers perform poorly in coordination construction (i.e., correctly attaching the "conj" relation). We extend a state-of-the-art dependency parser with conjunction-specific features, focusing on the similarity between the conjuncts head words. Training the extended parser yields an improvement in "conj" attachment as well as in overall dependency parsing accuracy on the Stanford dependency conversion of the Penn TreeBank.