English (United Kingdom)

https://curated-unify.zendy.io/wp-json/zendy-region/v1/featured_content/oa?rat=en

https://curated-unify.zendy.io/wp-json/zendy-region/v1/highlighted_journal/

Global: Zendy Plus annual

Presents the access of premium content as premium feature

Premium Content

Presents the keyphrase highlighting as premium feature

Keyphrase Highlighting

Presents the summarisation as premium feature

Summarisation

Insights

Presents the pdf analysis as premium feature

PDF Analysis

Presents the zaia usage as premium feature

ZAIA

Global: Zendy Plus monthly

Global: Zendy Tools annual

Global: Zendy Tools monthly

Global: Zendy Free Plan

Biomedical relation extraction (RE) is the task of automatically identifyingand characterizing relations between biomedical concepts from free text. RE isa central task in biomedical natural language processing (NLP) research andplays a critical role in many downstream applications, such as literature-baseddiscovery and knowledge graph construction. State-of-the-art methods were usedprimarily to train machine learning models on individual RE datasets, such asprotein-protein interaction and chemical-induced disease relation. Manualdataset annotation, however, is highly expensive and time-consuming, as itrequires domain knowledge. Existing RE datasets are usually domain-specific orsmall, which limits the development of generalized and high-performing REmodels. In this work, we present a novel framework for systematicallyaddressing the data heterogeneity of individual datasets and combining theminto a large dataset. Based on the framework and dataset, we report on BioREx,a data-centric approach for extracting relations. Our evaluation shows thatBioREx achieves significantly higher performance than the benchmark systemtrained on the individual dataset, setting a new SOTA from 74.4% to 79.6% inF-1 measure on the recently released BioRED corpus. We further demonstrate thatthe combined dataset can improve performance for five different RE tasks. Inaddition, we show that on average BioREx compares favorably to currentbest-performing methods such as transfer learning and multi-task learning.Finally, we demonstrate BioREx's robustness and generalizability in twoindependent RE tasks not previously seen in training data: drug-drug N-arycombination and document-level gene-disease RE. The integrated dataset andoptimized method have been packaged as a stand-alone tool available athttps://github.com/ncbi/BioREx.

BioREx: Improving Biomedical Relation Extraction by Leveraging Heterogeneous Datasets