A deep learning relation extraction approach to support a biomedical semi-automatic curation task: The case of the gluten bibliome

•A semi-automatic workflow to curate health-related interactions in the bibliome.•Different text minning an ontology-based methods were combined and applied.•A deep learning model to assist manual curators learning from their decisions.•Different state-of-the-art machine learning models were compared.•Results showed that the presented workflow is a valuable approach to assist curators.