VigiCovid :

Search engines and knowledge management

Overview

We created VIGICOVID, a system to automatically extract information from scientific articles on COVID-19. By asking questions in natural language it is a system designed to get answers from the deluge of information dealing with COVID-19 and SARS-CoV-2.  

The research conducted worldwide on the pandemic has produced a deluge of information, so the system addressed the need to find answers in all this information. Through AI and natural language answers and questions, it presents answers to the user in an organised way from a collection comprising scientific publications.

Challenge

Biohealth researchers across the globe are making a huge effort in generating knowledge about COVID-19 and SARS-CoV-2. As a result of that effort, multiple scientific publications are being produced very quickly, and that hampers being able to consult and analyse all the information. This project explored innovative neural perspectives to enable these huge volumes of information to be consulted in natural language.

Collaboration

We worked in collaboration with researchers of the HiTZ Centre at the UPV/EHU-University of the Basque Country, the NLP & IR group of the UNED (Distance University) and Orai NLP Technology, taking advantage of  Fondo Supera COVID-19 funding provided by the CRUE, the CSIC (National Research Council) and the Banco Santander. 

Result

http://nlp.uned.es/vigicovid-project/index.html 

Arantxa Otegi, Iñaki San Vicente, Xabier Saralegi, Anselmo Peñas, Borja Lozano, Eneko Agirre: 
Information retrieval and question answering: A case study on COVID-19 scientific literature Knowledge-Based Systems DOI: 10.1016/j.knosys.2021.108072