Welcome To SemanticCORD-19 Search

Search over documents from COVID-19 Open Research Dataset dataset.
This project was started to work on semantic analysis of papers and contribute to ongoing efforts against COVID-19.
Check Swagger API how access to documents and perform LDA feature extraction analysis of included texts.

Who is this site for
Use this site to quickly find related documents and extracted features. It was built because sometimes it's hard to use commercial search services when looking in domain specific knowledge. When Allen Institute for AI published CORD-19 Open Research dataset I decided i will do my part and created this site. In order to view complete document in PDF format your will have to find them on corresponding archives.

Some sample querys and how to get excerpts from literature

Search for specific words

Get results based on paper features and topics

Mind this project is not yet feature complete.
  • Linking of documents by citation
  • No figures nor tables included
  • No Data visualization
  • Search query options are not final
  • no results scoring
  • query order matters, ie. query '-animal +immune' will remove all papers with 'animal' before 'immune' search

Other sources for reference (not included here)

Genome

How this was project made
  • All documents were combined
  • LDA prediction engine was trained
  • Topics and feature values were predicted for each document
  • Each paper contains topics extracted from its text.
  • App to search papers was written and API to access its values and use prediction engine
  • This project is a memory hog, runs on machine and eats approximately 5 to 8 GB of space

This site serves CORD-19 documents from CORD-19 enriched with extracted topics and features
By using this site you agree to Dataset License
Please see additional licensing information PMC website, medRxiv and bioRxiv

COVID-19 Open Research Dataset (CORD-19, 2020). Version 2020-03-13. Retrieved from https://pages.semanticscholar.org/coronavirus-research. Accessed 2020-03-14. doi:10.5281/zenodo.3715506