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 Tools annual

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 Tools monthly

Global: Zendy Plus monthly

Presents the access of premium content as premium feature

Premium Content

Global: Zendy Plus annual

Global: Zendy Free Plan

In this paper we address the challenge of extracting scientific referencesfrom patents. We approach the problem as a sequence labelling task andinvestigate the merits of BERT models to the extraction of these longsequences. References in patents to scientific literature are relevant to studythe connection between science and industry. Most prior work only uses thefront-page citations for this analysis, which are provided in the metadata ofpatent archives. In this paper we build on prior work using Conditional RandomFields (CRF) and Flair for reference extraction. We improve the quality of thetraining data and train three BERT-based models on the labelled data (BERT,bioBERT, sciBERT). We find that the improved training data leads to a largeimprovement in the quality of the trained models. In addition, the BERT modelsbeat CRF and Flair, with recall scores around 97% obtained with crossvalidation. With the best model we label a large collection of 33 thousandpatents, extract the citations, and match them to publications in the Web ofScience database. We extract 50% more references than with the old trainingdata and methods: 735 thousand references in total. With thesepatent-publication links, follow-up research will further analyze which typesof scientific work lead to inventions.

Improving reference mining in patents with BERT