Article Preview
TopIntroduction
Covid-19 an ongoing global pandemic disease is caused by a novel coronavirus. It has sensationally affected human life in the entire world and given challenges to the global health system, societal system, economy, work culture etc. Many countries have applied lockdown to prevent the spread of coronavirus disease. Lockdown smashes the monitory and social disturbance among people as they have been locked in their homes and can move only for essential services. Many of the studies reported the psychological impact of lockdown on human behavior like anger, sorrow, depression, frustration etc. (Balhara et al., 2020; Kantermann, 2020; Kumar & Dwivedi, 2020; Prati & Mancini, 2021; Bera et al., 2021). All these behaviors adversely impact the health of a human being. The dynamics of Covid-19 comprising mortalities, number of daily cases, number of active cases, aftereffects of coronavirus on patients, impact on children etc. has been shared by people on the social media during severe pandemic (Hussain, 2020; Goel & Gupta, 2020; Gao et al., 2020).
The current pandemic has changed the way of living, behaving, and communicating with each other all around the world (Hung et al., 2020; Zang, 2021). These specific circumstances have raised the challenges of movement of data while educating, working, and communicating online in the public arena. The pandemic due to Covid-19 has also brought a genuine challenge for governments, businessmen, associations, and media to improve the existing systems to control the virus spread. The current crisis is globally forming a socially advanced situation due to lock on the free movements during relocation, travel among states, travel among nations, number of gatherings in events and many more. Now, every movement of people is dependent on Information and Communication Technology (ICT).
ICT has supported people to communicate through social media platforms to share their thoughts and emotions during this corona outbreak. Twitter is one of the social media platforms which is majorly used by people for distributing and getting information. Plenty of information is available on twitter in the form of tweets. By analyzing these tweets one can determine the state of mind (happy, bad, depressed, frustrated etc.) of people during lockdown. By knowing the state of mind of people, government and other social groups can help them to bear the pandemic. In this work, we have considered the large dataset of tweets for sentiment analysis on textual data on Covid-19. In open literature, some of the studies suggested making use of machine learning techniques to improve the performance of sentiment analysis. As per our knowledge, no one has applied the multinomial logistic regression algorithm for performing the sentiment analysis on Covid-19 data. However, the technique is used for other datasets. This study aims to propose a framework using multinomial logistic regression to perform sentiment analysis on Covid-19 twitter data.
Rest of the paper was organized as follows: Section 2 discusses the background and literature review. Section 3 introduces the proposed system framework adopted for the current work. Section 4 discusses the multinomial logistic regression algorithm in detail followed by result and discussion in section 5. Section 6 concludes the present study followed by future work.
Acronyms Used
BERT: Bidirectional Encoder Representations from Transformers
LR: logistic regression
SVM: Support Vector Machines
LSTM: Long-short Term Memory
NLP: Natural Language Processing
RF: Random Forest
ML: Machine Learning