Explore BiLSTM-CRF-Based Models for Open Relation Extraction

TL;DR

This paper develops BiLSTM-CRF-based models with novel tagging schemes and contextualized embeddings to improve open relation extraction from text, especially for sentences with multiple relations.

Contribution

It introduces a new tagging scheme and evaluates various embedding methods to enhance the performance of open relation extraction models.

Findings

Best model achieves high accuracy on multiple-relation sentences

New tagging scheme effectively resolves overlapping relation issues

Model outperforms existing approaches in open relation extraction

Abstract

Extracting multiple relations from text sentences is still a challenge for current Open Relation Extraction (Open RE) tasks. In this paper, we develop several Open RE models based on the bidirectional LSTM-CRF (BiLSTM-CRF) neural network and different contextualized word embedding methods. We also propose a new tagging scheme to solve overlapping problems and enhance models' performance. From the evaluation results and comparisons between models, we select the best combination of tagging scheme, word embedder, and BiLSTM-CRF network to achieve an Open RE model with a remarkable extracting ability on multiple-relation sentences.