COVID-19 using LDA topic modeling
Loading...
Date
2021
Authors
Journal Title
Journal ISSN
Volume Title
Publisher
UMT, Lahore
Abstract
Nowadays COVID-19 is a trending matter, there are number of publication on COVID 19. Most of the papers based on sentiment analysis. A vast number of study issued on COVID-19. Consequently, in this study researcher present an academic construction dataset positioned on COVID-19, which gives a suggestion of numerous method actions and allows people to examine the states, geniuses, and study of organizations that are most affianced in the coronavirus task force. Goal of this research is to scrape data from different articles through PubMed then scraped data have to preprocess for model implementation. The data meticulously gathered, with almost 10 thousand publications from a total satisfaction of chosen to provide the most inclusive information sources to cover the research issues. The main contribution is to look at the most frequent topics throughout the study using Latent Dirichlet Allocation (LDA). In this research, two models utilized both models yielded satisfactory results. Affinity propagation used to tune the LDA model. This produced a total number of seven topics, which explains 95% of the information and keeping a high perplexity and coherence rate. In the future, researcher will be able to use effective models to provide precise results.