GU IRLAB at SemEval-2018 Task 7: Tree-LSTMs for Scientific Relation Classification

TL;DR

This paper presents a tree-based LSTM model for scientific relation classification in literature, achieving top-10 results in SemEval 2018 tasks and analyzing feature contributions.

Contribution

The work introduces a novel tree-LSTM approach for scientific relation classification and provides an ablation study of input features.

Findings

Placed 9th out of 28 in subtask 1.1

Placed 5th out of 20 in subtask 1.2

Conducted feature ablation analysis

Abstract

SemEval 2018 Task 7 focuses on relation ex- traction and classification in scientific literature. In this work, we present our tree-based LSTM network for this shared task. Our approach placed 9th (of 28) for subtask 1.1 (relation classification), and 5th (of 20) for subtask 1.2 (relation classification with noisy entities). We also provide an ablation study of features included as input to the network.