Building a Relation Extraction Baseline for Gene-Disease Associations: A Reproducibility Study

TL;DR

This paper reproduces the DEXTER system for extracting gene-disease associations from biomedical texts to establish a reproducible benchmark for relation extraction research.

Contribution

It provides a reproducibility study of DEXTER, creating a benchmark dataset for future gene-disease relation extraction research.

Findings

Reproduced DEXTER system successfully

Established a benchmark for GDA relation extraction

Facilitates comparison of future methods

Abstract

Reproducibility is an important task in scientific research. It is crucial for researchers to compare newly developed systems with the state-of-the-art to assess whether they made a breakthrough. However previous works may not be immediately reproducible, for example due to the lack of source code. In this work we reproduce DEXTER, a system to automatically extract Gene-Disease Associations (GDAs) from biomedical abstracts. The goal is to provide a benchmark for future works regarding Relation Extraction (RE), enabling researchers to test and compare their results.