International Research Journal of Engineering and Technology (IRJET) Volume: 09 Issue: 05 | May 2022
e-ISSN: 2395-0056
www.irjet.net
p-ISSN: 2395-0072
COVID Sentiment Analysis of Social Media Data Using Enhanced Stacked Ensemble E. Prabhakar1, P.S Aiswarya2, A. Jamuna Banu3, M. Kowsalya4 ,P. Kanimozhi5 1Assistant
Professor, Dept. of Computer Science and Engineering(CSE), Nandha College of Technology, TamilNadu, India 2,3,4,5Final Year CSE, Dept. of CSE, Nandha College of Technology, TamilNadu, India ---------------------------------------------------------------------***---------------------------------------------------------------------
Abstract - COVID-19 triggered a global public health crisis and a slew of other concerns, including an economic downturn, job
losses, and mental anguish. This epidemic affects people all around the world, causing anxiety, stress, worry, fear, repugnance, and poignancy in addition to the sickness. During this time, social media involvement and interaction skyrocketed, allowing people to share their thoughts and opinions on the aforementioned health issues. We can assess the people's ideas and attitudes on health status, concerns, panic, and awareness related to social issues using user-generated content on social media, which can help establish health intervention techniques and plan effective campaigns based on popular impressions. On the COVID-19 tweets dataset, we look at user sentiment at different time intervals to help with hot topics on Twitter. This research contributes a novel way of analysing social media on Twitter. It provides new information on the impact of sentiment polarity in COVID-19 tweets on tweet responses. Key Words: Sentiment Analysis, COVID-19, Twitter Data, Stacked Ensemble, Public Opinion, Social Media Data.
1. INTRODUCTION Twitter, for example, has gone a mainstream venue for numerous people to share their opinions on current events as a series of tweets [1]. It has a large, global impact on popular perceptions of current events [2, 3]. The pandemic's illness burden has unavoidable consequences for population health and well-being, health resource use, social dynamics, global economies, and health technology development. Stakeholders from all around the world (governments, non-profit organizations, and healthcare groups) have been working hard to combat the COVID-19 outbreak. Surprisingly, the COVID-19 pandemic has sparked a flood of brief informal texts on Twitter, reflecting public fears, concerns, and a new type of discrimination [4]. As a result, it's never been more important to comprehend and draw conclusions from the vast number of social media messages, such as Tweets. This knowledge will aid in the development of public health campaigns or social media events to refute misinformation and combat the global fear and stigma surrounding COVID-19. Our effort covers the concept of assessing active users' various attitudes, such as positive, negative, and neutral sentiments, toward trending topics linked to COVID-19 at a specific time interval. This study focuses on people's positive, negative, and neutral feelings on COVID-19's top-k trending sub-topics on Twitter. The primary contributions of our research include Evaluating the effectiveness of existing algorithms and proposing a model that successfully classifies emotions. The remainder of the paper is laid out as follows. Related work is presented in Section 2. The proposed mechanism for determining feelings is detailed in Section 3. Section 4 discusses the findings of the experiments. Finally, Section 5 brings the work to close-by outlining future research prospects.
2. LITERATURE SURVEY By using the power of Natural language processing (NLP) to analyze the sentiment that is being transmitted in the particular data, the notion of opinion mining or sentiment analysis has been employed and developed for diverse analyses over time [5]. This section summarizes past empirical research in this field. The BERT model is used in the paper [6] to perform Sentiment Analysis on Twitter data. The location of individual tweets was utilized to categorize the data presented in this article. The BERT model for emotion categorization was used to train the data, and the model's performance was assessed using the SVM classifier. Using COVID-19 Twitter data, a model for assessing the influence of COVID-19 on stocks was developed in the article [7]. With an accuracy of 86.24 percent, this model was trained using supervised learning. The goal of this study was to assist businesses in
© 2022, IRJET
|
Impact Factor value: 7.529
|
ISO 9001:2008 Certified Journal
|
Page 2495