End-to-End Relation Extraction using LSTMs on Sequences and Tree Structures

TL;DR

This paper introduces a novel neural network model combining sequence and tree-structured LSTMs for end-to-end relation extraction, achieving significant improvements over previous methods on benchmark datasets.

Contribution

The authors propose a unified LSTM-based model that jointly captures entity and relation information using sequence and dependency tree structures, with training strategies to enhance entity detection.

Findings

Achieved 12.1% and 5.7% relative error reductions in F1-score on ACE datasets.

Outperformed state-of-the-art feature-based models in end-to-end relation extraction.

Compared favorably to CNN-based models on nominal relation classification.

Abstract

We present a novel end-to-end neural model to extract entities and relations between them. Our recurrent neural network based model captures both word sequence and dependency tree substructure information by stacking bidirectional tree-structured LSTM-RNNs on bidirectional sequential LSTM-RNNs. This allows our model to jointly represent both entities and relations with shared parameters in a single model. We further encourage detection of entities during training and use of entity information in relation extraction via entity pretraining and scheduled sampling. Our model improves over the state-of-the-art feature-based model on end-to-end relation extraction, achieving 12.1% and 5.7% relative error reductions in F1-score on ACE2005 and ACE2004, respectively. We also show that our LSTM-RNN based model compares favorably to the state-of-the-art CNN based model (in F1-score) on nominal relation classification (SemEval-2010 Task 8). Finally, we present an extensive ablation analysis of several model components.