Building a reverse dictionary with specific application to the COVID-19 pandemic.

Abstract

A Reverse Dictionary maps a natural language description to corresponding semantically appropriate words. It is of assistance, particularly to the language producers, in finding the correct word for a concept in mind while writing/speaking. As the COVID-19 pandemic intensely impacted almost all the functionalities across the globe, texts on this subject appear in a significant amount in various forms, including news updates, awareness and safety articles, notices and circulars, research articles, social media posts, etc. A Reverse Dictionary on this subject is a requisite in view of the following reasons, hence addressed. Firstly, the varied text forms involve a diverse range of language producers ranging from professional doctors to the general mass. Secondly, the COVID-19 pandemic’s glossary is more specific than the general English language, hence unfamiliar to the language producers. We have carried out an implementation based on the Wordster Reverse Dictionary architecture, owing to its outperformance of the commercial Onelook Reverse Dictionary benchmark. We report an accuracy of 0.49 based on top-3 system responses. To address the limitations of the current implementation, we bring into consideration Zadeh’s paradigm of the Computational Theory of Perceptions. Notably, the compilation of the COVID-19 glossary as a part of this study is another contribution in view that it is of assistance to the concerned readers.