Fig. 3From: Assisted neuroscience knowledge extraction via machine learning applied to neural reconstruction metadata on NeuroMorpho.OrgOverview of the architecture of the metadata extraction system used in this study. The full text of the publication is preprocessed, marked with sentence boundaries (CLS and SEP), and converted into sequences of words. Each sequence is then passed through the entity extraction algorithm to identify occurrences of metadata entities. The candidate terms are then post-processed and ranked based on relevance scoreBack to article page