%0 Journal Article
%9 ACL : Articles dans des revues avec comité de lecture répertoriées par l'AERES
%A Nainia, A.
%A Vignes-Lebbe, R.
%A Chenin, Eric
%A Sahraoui, M.
%A Mousannif, H.
%A Zahir, J.
%T FloraNER : a new dataset for species and morphological terms named entity recognition in French botanical text
%D 2024
%L fdi:010091277
%G ENG
%J Data in Brief
%@ 2352-3409
%K NER Dataset ; Biodiversity dataset ; Species identification dataset ; Plant morphology dataset
%K NOUVELLE CALEDONIE
%M ISI:001299243700001
%P 110824 [10 ]
%R 10.1016/j.dib.2024.110824
%U https://www.documentation.ird.fr/hor/fdi:010091277
%> https://horizon.documentation.ird.fr/exl-doc/pleins_textes/2024-10/010091277.pdf
%V 56
%W Horizon (IRD)
%X FloraNER is a distantly supervised named entity recognition dataset (NER). The dataset is built from botanical French literature extracted from the OCR-preprocessed flora of New Caledonia, provided by the National Museum of Natural History in France (MNHN), and distantly annotated with a botanical French corpus created by merging botanical lexicons available online. FloraNER comprises separate subdatasets for the recognition of plant species names, as well as coarse-grained and fine-grained botanical morphological terms. The resulting datasets are in CSV format, displaying textual data, identified named entities, and their annotations, covering one named entity type "Species" (Esp & egrave;ce in French) for species name identification, two named entity types "Organ" and "Descriptor" for coarse-grained morphological term identification, and eight named entity types for fine-grained morphological term identification: Organ, Descriptor, Form, Color, Development, Structure, Surface, Position, Disposition, and Measure. This dataset can be utilized to train and evaluate named entity recognition models for extracting information from botanical French literature.
%$ 076 ; 124