INKAD COVID-19 IntelliSearch: a multilingual search engine for answering questions about COVID-19 in real-time from the scientific literature
Zakaria Kaddari, Jamal Berrich, Noureddine Rahmoun, Saïda Belouali, Toumi Bouchentouf · 2021
Thousands of research papers on COVID-19 have been published since the start of the pandemic. To find relevant information in this vast literature, researchers and healthcare information professionals, spend increasingly more time per search query. In this paper, we present INKAD COVID-19 IntelliSearch, a multilingual search engine that we built to help researchers and healthcare information professionals in finding precise and relevant information from the COVID-19 literature in real-time, while considerably reducing time spent per search query. We used the COVID-19 Open Research Dataset as the main source of papers. The search engine has a BM25 based document retrieval component, and a neural question-answering component returning the exact answer span. The overall system is evaluated against a COVID-19 question-answering test set with different information retrieval and question-answering models. We have made INKAD COVID-19 IntelliSearch accessible online for broader use by researchers and medical information professionals.