Get our free extension to see links to code for papers anywhere online!

Chrome logo  Add to Chrome

Firefox logo Add to Firefox

Zero-Shot and Few-Shot Classification of Biomedical Articles in Context of the COVID-19 Pandemic

Jan 11, 2022
Simon Lupart, Benoit Favre, Vassilina Nikoulina, Salah Ait-Mokhtar

Share this with someone who'll enjoy it:

MeSH (Medical Subject Headings) is a large thesaurus created by the National Library of Medicine and used for fine-grained indexing of publications in the biomedical domain. In the context of the COVID-19 pandemic, MeSH descriptors have emerged in relation to articles published on the corresponding topic. Zero-shot classification is an adequate response for timely labeling of the stream of papers with MeSH categories. In this work, we hypothesise that rich semantic information available in MeSH has potential to improve BioBERT representations and make them more suitable for zero-shot/few-shot tasks. We frame the problem as determining if MeSH term definitions, concatenated with paper abstracts are valid instances or not, and leverage multi-task learning to induce the MeSH hierarchy in the representations thanks to a seq2seq task. Results establish a baseline on the MedLine and LitCovid datasets, and probing shows that the resulting representations convey the hierarchical relations present in MeSH.

* to be published at the AAAI-22 Workshop on Scientific Document Understanding 

   Access Paper Source

Share this with someone who'll enjoy it: