Structural block driven - enhanced convolutional neural representation for relation extraction

TL;DR

This paper introduces a lightweight, block-driven CNN approach for relation extraction that focuses on essential token sequences to improve accuracy and reduce noise, achieving state-of-the-art results on KBP37.

Contribution

The method uniquely uses dependency analysis to identify relevant token blocks and encodes them with multi-scale CNNs, enhancing relation extraction performance.

Findings

Achieves new state-of-the-art on KBP37 dataset.

Performs comparably to top methods on SemEval2010.

Effectively reduces noise by focusing on relevant token blocks.

Abstract

In this paper, we propose a novel lightweight relation extraction approach of structural block driven - convolutional neural learning. Specifically, we detect the essential sequential tokens associated with entities through dependency analysis, named as a structural block, and only encode the block on a block-wise and an inter-block-wise representation, utilizing multi-scale CNNs. This is to 1) eliminate the noisy from irrelevant part of a sentence; meanwhile 2) enhance the relevant block representation with both block-wise and inter-block-wise semantically enriched representation. Our method has the advantage of being independent of long sentence context since we only encode the sequential tokens within a block boundary. Experiments on two datasets i.e., SemEval2010 and KBP37, demonstrate the significant advantages of our method. In particular, we achieve the new state-of-the-art performance on the KBP37 dataset; and comparable performance with the state-of-the-art on the SemEval2010 dataset.