Using AI to build automated glossaries of COVID-19 research papers

CSER Researcher, Dr Thushari Atapattu and PhD student Menasha Thilakaratne, were successfully awarded $20,000 from the South Australian Defence Innovation Partnership Program to commence research on developing an ‘automated glossary from scientific articles on COVID-19’ under the initiative on ‘AI for decision making’.

In novel areas of interest (e.g. COVID-19), extracting the right information for quick and informed decision-making is an extremely challenging task due to the domain expertise needed and the rapid growth of information available. With an overwhelming abundance of information on a selected topic, glossaries serve as a bridge between the information-seeker and the information itself. The glossary creation process is inherently complex and depends largely on a domain expert's knowledge and with the abundant and ever-evolving nature of knowledge emerging within scientific domains, manual glossary creation is very labour-intensive.

Our researchers will be using Artificial Intelligence (AI) and in particular novel Natural Language Processing (NLP) techniques to automatically generate glossaries from a large corpus of scientific articles on COVID-19 in response to user queries. The techniques developed in this project, within the context of COVID-19, are advancing novel AI techniques and applications and can be explored across a whole host of topics and for different purposes.