Structured Minimally Supervised Learning for Neural Relation Extraction

TL;DR

This paper introduces a structured minimally supervised learning method for neural relation extraction that effectively handles label noise and achieves state-of-the-art results using proposition-level supervision from knowledge bases.

Contribution

It combines structured learning with neural networks to improve relation extraction accuracy under minimal supervision, explicitly reasoning about missing data.

Findings

Achieved state-of-the-art results on minimally supervised relation extraction.

Effectively mitigated label noise in distant supervision.

Enabled large-scale training of CNNs with improved accuracy.

Abstract

We present an approach to minimally supervised relation extraction that combines the benefits of learned representations and structured learning, and accurately predicts sentence-level relation mentions given only proposition-level supervision from a KB. By explicitly reasoning about missing data during learning, our approach enables large-scale training of 1D convolutional neural networks while mitigating the issue of label noise inherent in distant supervision. Our approach achieves state-of-the-art results on minimally supervised sentential relation extraction, outperforming a number of baselines, including a competitive approach that uses the attention layer of a purely neural model.