Position-aware Self-attention with Relative Positional Encodings for Slot Filling

TL;DR

This paper introduces a position-aware self-attention model with relative positional encodings for relation extraction, achieving state-of-the-art results on TACRED without using recurrent or convolutional layers.

Contribution

It proposes a novel self-attention architecture with relative positional encodings and a position-aware attention layer for improved relation extraction.

Findings

Achieved state-of-the-art performance on TACRED dataset.

Model relies solely on attention mechanisms, no RNNs or CNNs used.

Improved relation extraction accuracy over previous methods.

Abstract

This paper describes how to apply self-attention with relative positional encodings to the task of relation extraction. We propose to use the self-attention encoder layer together with an additional position-aware attention layer that takes into account positions of the query and the object in the sentence. The self-attention encoder also uses a custom implementation of relative positional encodings which allow each word in the sentence to take into account its left and right context. The evaluation of the model is done on the TACRED dataset. The proposed model relies only on attention (no recurrent or convolutional layers are used), while improving performance w.r.t. the previous state of the art.