Syntactic Dependency Representations in Neural Relation Classification

TL;DR

This paper explores how various syntactic dependency representations impact neural relation classification performance, comparing multiple schemes and a syntax-agnostic approach to understand their effectiveness.

Contribution

It provides a comparative analysis of different dependency schemes and introduces insights into their influence on neural relation classification accuracy.

Findings

Dependency scheme choice affects classification performance

Universal Dependencies scheme shows competitive results

Error analysis reveals specific cases where syntax helps or hinders

Abstract

We investigate the use of different syntactic dependency representations in a neural relation classification task and compare the CoNLL, Stanford Basic and Universal Dependencies schemes. We further compare with a syntax-agnostic approach and perform an error analysis in order to gain a better understanding of the results.