A Sentence Simplification System for Improving Relation Extraction

TL;DR

This paper introduces a sentence simplification system designed to enhance the accuracy of open relation extraction by preprocessing complex sentences into simpler forms using syntactic transformation rules.

Contribution

It proposes a novel text simplification framework specifically aimed at improving relation extraction performance in complex sentences.

Findings

Improved relation extraction accuracy on complex sentences

Effective syntactic transformation rules for simplification

Preprocessing enhances downstream relation extraction systems

Abstract

In this demo paper, we present a text simplification approach that is directed at improving the performance of state-of-the-art Open Relation Extraction (RE) systems. As syntactically complex sentences often pose a challenge for current Open RE approaches, we have developed a simplification framework that performs a pre-processing step by taking a single sentence as input and using a set of syntactic-based transformation rules to create a textual input that is easier to process for subsequently applied Open RE systems.