Multimodal negative sentiment recognition of online public opinion on public health emergencies based on graph convolutional networks and ensemble learning

To improve the effect of multimodal negative sentiment recognition of online public opinion on public health emergencies, we constructed a novel multimodal fine-grained negative sentiment recognition model based on graph convolutional networks (GCN) and ensemble learning. This model comprises BERT and ViT-based multimodal feature representation, GCN-based feature fusion, multiple classifiers, and ensemble learning-based decision fusion. Firstly, the image-text data about COVID-19 is collected from Sina Weibo, and the text and image features are extracted through BERT and ViT, respectively. Secondly, the image-text fused features are generated through GCN in the constructed microblog graph. Finally, AdaBoost is trained to decide the final sentiments recognized by the best classifiers in image, text, and image-text fused features. The results show that the F1-score of this model is 84.13% in sentiment polarity recognition and 82.06% in fine-grained negative sentiment recognition, improved by 4.13% and 7.55% compared to the optimal recognition effect of image-text feature fusion, respectively. © 2023 Elsevier Ltd

Do SEC filings indicate any trends? Evidence from the sentiment distribution of forms 10-K and 10-Q with FinBERT

Purpose: This study quantified companies' views on the COVID-19 pandemic with sentiment analysis of US public companies' disclosures. The study aims to provide timely insights to shareholders, investors and consumers by exploring sentiment trends and changes in the industry and the relationship with stock price indices. Design/methodology/approach: From more than 50,000 Form 10-K and Form 10-Q published between 2020 and 2021, over one million texts related to the COVID-19 pandemic were extracted. Applying the FinBERT fine-tuned for this study, the texts were classified into positive, negative and neutral sentiments. The correlations between sentiment trends, differences in sentiment distribution by industry and stock price indices were investigated by statistically testing the changes and distribution of quantified sentiments. Findings: First, there were quantitative changes in texts related to the COVID-19 pandemic in the US companies' disclosures. In addition, the changes in the trend of positive and negative sentiments were found. Second, industry patterns of positive and negative sentiment changes were similar, but no similarities were found in neutral sentiments. Third, in analyzing the relationship between the representative US stock indices and the sentiment trends, the results indicated a positive relationship with positive sentiments and a negative relationship with negative sentiments. Originality/value: Performing sentiment analysis on formal documents like Securities and Exchange Commission (SEC) filings, this study was differentiated from previous studies by revealing the quantitative changes of sentiment implied in the documents and the trend over time. Moreover, an appropriate data preprocessing procedure and analysis method were presented for the time-series analysis of the SEC filings. © 2022, Emerald Publishing Limited.

The Data Leakage Sentiment Analysis Using Naive Bayes Algorithm Based on Machine Learning Approach

Data leakage is a case that often occurs anywhere. Indonesia is one of the countries with the most population that is currently having data leakage cases. The leak of data on the COVID-19 PeduliLindung tracking application, triggered a public reaction because it was considered dangerous. Based on this, the aim of the study is to predict the sentiment pattern using Naïve Bayes. This study is important to do sentiment analysis to find out the public's reaction, it can become a recommendation in developing applications that are safer in data storage. The experiment in this study used data from Twitter which was taken for 14 days, between 16-21 May 2022. The data was processed using Google Collab and the Naïve Bayes approach. The experimental results are that negative sentiment is greater than positive sentiment and neutral sentiment, which is 93%. While the accuracy of positive sentiment is 81% and Neutra sentiment is 90%. This means the leak of public data from a COVID-19 tracing application in Indonesia has a greater negative sentiment. The difference in the study is the data testing process was carried out five times to get good accuracy from the model. And the results show that Naïve Bayes is a model that is quite recommended for prediction of sentiment patterns. © 2022 IEEE.

Sentimental Analysis of COVID-19 Vaccine Tweets Using BERT+NBSVM

The development of the vaccine for the control of COVID-19 is the need of hour. The immunity against coronavirus highly depends upon the vaccine distribution. Unfortunately, vaccine hesitancy seems to be another big challenge worldwide. Therefore, it is necessary to analysis and figure out the public opinion about COVID-19 vaccines. In this era of social media, people use such platforms and post about their opinion, reviews etc. In this research, we proposed BERT+NBSVM model for the sentimental analysis of COVID-19 vaccines tweets. The polarity of the tweets was found using TextBlob(). The proposed BERT+NBSVM outperformed other models and achieved 73% accuracy, 71% precision, 88% recall and 73% F-measure for classification of positive sentiments while 73% accuracy, 71% precision, 74% recall and 73% F-measure for classification of negative sentiments respectively. Thus, these sentimental and spatial analysis helps in world-wide pandemics by identify the people's attitudes towards the vaccines. © 2023, The Author(s), under exclusive license to Springer Nature Switzerland AG.

Network distribution and sentiment interaction: Information diffusion mechanisms between social bots and human users on social media

When public health emergencies occur, a large amount of low-credibility information is widely disseminated by social bots, and public sentiment is easily manipulated by social bots, which may pose a potential threat to the public opinion ecology of social media. Therefore, exploring how social bots affect the mechanism of information diffusion in social networks is a key strategy for network governance. This study combines machine learning methods and causal regression methods to explore how social bots influence information diffusion in social networks with theoretical support. Specifically, combining stakeholder perspective and emotional contagion theory, we proposed several questions and hypotheses to investigate the influence of social bots. Then, the study obtained 144,314 pieces of public opinion data related to COVID-19 in J city from March 1, 2022, to April 18, 2022, on Weibo, and selected 185,782 pieces of data related to the outbreak of COVID-19 in X city from December 9, 2021, to January 10, 2022, as supplement and verification. A comparative analysis of different data sets revealed the following findings. Firstly, through the STM topic model, it is found that some topics posted by social bots are significantly different from those posted by humans, and social bots play an important role in certain topics. Secondly, based on regression analysis, the study found that social bots tend to transmit information with negative sentiments more than positive sentiments. Thirdly, the study verifies the specific distribution of social bots in sentimental transmission through network analysis and finds that social bots are weaker than human users in the ability to spread negative sentiments. Finally, the Granger causality test is used to confirm that the sentiments of humans and bots can predict each other in time series. The results provide practical suggestions for emergency management under sudden public opinion and provide a useful reference for the identification and analysis of social bots, which is conducive to the maintenance of network security and the stability of social order. © 2022

A global portrait of expressed mental health signals towards COVID-19 in social media space.

Globally, the COVID-19 pandemic has induced a mental health crisis. Social media data offer a unique opportunity to track the mental health signals of a given population and quantify their negativity towards COVID-19. To date, however, we know little about how negative sentiments differ across countries and how these relate to the shifting policy landscape experienced through the pandemic. Using 2.1 billion individual-level geotagged tweets posted between 1 February 2020 and 31 March 2021, we track, monitor and map the shifts in negativity across 217 countries and unpack its relationship with COVID-19 policies. Findings reveal that there are important geographic, demographic, and socioeconomic disparities of negativity across continents, different levels of a nation's income, population density, and the level of COVID-19 infection. Countries with more stringent policies were associated with lower levels of negativity, a relationship that weakened in later phases of the pandemic. This study provides the first global and multilingual evaluation of the public's real-time mental health signals to COVID-19 at a large spatial and temporal scale. We offer an empirical framework to monitor mental health signals globally, helping international authorizations, including the United Nations and World Health Organization, to design smart country-specific mental health initiatives in response to the ongoing pandemic and future public emergencies.

Customer Satisfaction Analysis of Mobile Banking Application Based on Twitter Data

The Covid-19 pandemic has changed the way people transact from physical bank to mobile banking transactions. The banks as financial companies are competing to offer the best service for mobile banking. Failure to fulfill consumer needs can damage a bank's reputation, profitability, and lead to gradual loss of customers. Hence, the bank needs to know its performance by measuring customer satisfaction. This study is aimed to know customer satisfaction of mobile banking using sentiment analysis. Sentiment analysis was conducted using data from Twitter, knowing that Twitter is a widely used media social with text-based content. The collected data was then predicted using the Support Vector Machine (SVM) model to get a positive or negative sentiment. From the model training result, this model achieved 92.5% of accuracy. The research analyzes customer satisfaction using sentiment analysis of BCA Mobile, Livin' by Mandiri, BRI Mobile (Brimo), and BNI Mobile. Brimo received the greatest percentage of positive sentiment compared to other platforms. Livin' by Mandiri and Brimo had a serious issue regarding its reliability. BCA Mobile should be more concerned regarding its usefulness. Meanwhile, BNI Mobile should have been more worried about its platform responsiveness. © 2022 IEEE.

Attitudes wards Online Learning ring COVID-19: A Cluster and Sentiment Analyses

This paper investigated the attitudes of 702 college students toward the implementation of fully online learning during the COVID-19 pandemic. Toward this goal, responses of the students were collected and analyzed through hierarchical cluster and sentiment analyses using the R software. Hierarchical cluster analysis revealed hopeful and apprehensive attitudes toward online learning. Advantages of online learning emerged as positive sentiments while challenges and their impact on mental health emerged as negative sentiments. It is concluded that online learning is a promising platform of learning provided that its shortcomings are addressed. Implications to teaching are offered. © 2022 The authors and IOS Press.

Sentiment Analysis of Nationwide Lockdown amid COVID 19: Evidence from Pakistan

A growing number of people are using tweets about the recent coronavirus epidemic of COVID-19 as a dataset to determine how worried people are in different parts of the world. This study attempts to uncover the key sentiments expressed by Twitter users regarding the COVID-19 epidemic by categorizing the tweets into positive and negative sentiments utilizing several resources (such as the Twitter search application programming interface (API), the Tweepy Python library, and the CSV excel database), as well as some predefined search terms ('#LockdownPakistan.'). We extracted the text of English language tweets from 28th March-1st May 2020. We have performed the sentiment analysis and classified the tweets in a binary class of positive and negative. Further, we used the word frequencies of single (unigrams), double (bigrams), and three words to examine the gathered tweets (tri-grams). According to our data, the majority of tweets express a positive attitude, with the word for lockdown COVID-19 appearing frequently. When looking at frequency analysis, the word 'family and time' stood out among the other words, which suggests that tweets were mostly optimistic and sentiments of defeating SARS-COV-2 prevail. People are determined to spend the lockdown in a good way. However, a few of the negative tweets, nevertheless, should serve as a warning for healthcare officials to make appropriate arrangements. Public health crisis responses are today complicated and highly synchronized both offline and online. Social media is a significant medium that gives people the chance to communicate with healthcare authorities directly. © 2022 IEEE.

Comparing Twitter Sentiment and United States COVID-19 Vaccination Rates

As more people use social media as a source of news and information, it is important to understand its impact on individual health decisions. This article compares the sentiment expressed in COVID-19 related tweets with national rates for first dose vaccinations as recorded by the Centers for Disease Control and Prevention. To conduct the study, the text from over 570,000 COVID-related tweets from January 2021 to December 2021 was captured. The tweets were segregated by month and Google Cloud's Natural Language API was used determine the sentiment in each tweet, with each post labeled as having positive, negative, or neutral sentiment. Overall, there was greater prevalence of negative sentiment as compared with positive sentiment during the period of review, with 45% of tweets negative, 33% positive and 22% neutral. The number of positive and negative tweets was more balanced in the early months of 2021 (when the vaccine was first available) and became decidedly more negative in the later part of the year, as misinformation about the vaccines spread prolifically on social media. This comparison of the tweet sentiment to first-time vaccine doses in the US shows that misinformation about vaccines on social media appears to have had an impact on behavior. Vaccine adoption declined significantly in the latter half of 2021, even as vaccines and information from public health officials regarding their efficacy became more available to the general public. These findings are validated by subsequent analysis of word usage by month, with positive comments about vaccines and vaccination in January through May coinciding with high vaccination rates, and a negative conversational shift to variants, increased deaths and suspicion about vaccine safety and effectiveness later in the year during a stagnation period in vaccinations. © 2022, The Author(s), under exclusive license to Springer Nature Switzerland AG.

Sentiment Analysis Twitter Based Lexicon and Multilayer Perceptron Algorithm

As the COVID-19 pandemic begins, the perception of online lectures according to students needs to be researched, to find out whether students have positive or negative sentiments regarding online lectures so far. Therefore, it is necessary to conduct research on sentiment analysis about online lectures taken according to student comments via tweets on the Twitter platform. The extracted tweets data will then be analyzed using machine learning to predict student sentiment about online lectures. The multilayer perceptron algorithm is used in research because it can solve non-linear problems well and is easy to implement without complicated parameter settings. However, multilayer perceptron is a supervised learning algorithm so it requires data that has been labeled/classified. So that to label the data of online lecture tweets, lexicon-based sentiment analysis is used. A total of 2,391 Indonesian-language tweets were successfully extracted. The results of the study using lexicon-based showed that as many as 63.9% gave negative sentiments towards online lectures, and 29% gave positive sentiments while the remaining 7.1% gave neutral sentiments. Meanwhile, the prediction ability of the multilayer perceptron algorithm for tweets data in this online lecture produces an accuracy of 71%. © 2022 IEEE.

Analysis of Google Play Store's Sentiment Review on Indonesia's P2P Fintech Platform

Nowadays there are so many mobile phone-based investment applications, ranging from mutual funds, stocks, and P2P lending. While these investment applications are gaining huge attraction among the general masses, sometimes selecting the right platform still becomes a hot issue. This research aimed to analyze the sentiment on P2P lending applications and to determine the user's response due to the increase in the number of funds distribution during the COVID-19 pandemic. By doing so, this research could give some insight into the new and existing user. Data was obtained through assessment reviews on the Play store platform for the P2P A, P2P B, and P2P C applications. Assessment reviews were classified by using a data mining approach, TF-IDF feature extraction, and Naïve-Bayes classification method. This research showed that P2P A got 77% positive sentiment and 23% negative sentiment, P2P B got 36% positive sentiment and 64% negative sentiment, and P2P C got 68% positive sentiment and 32% negative sentiment. From the results of the study, it was found that P2P A got better results than both P2P B and P2P C. those were 77% positive sentiment with 23% negative sentiment in finance topic, 56% positive sentiment with 44 % negative sentiment in account verification topic, 79% positive sentiment with 21% negative sentiment in apps review, and 99% positive sentiment with 1% negative sentiment in referral topic. © 2022 IEEE.

Are University Students Independent: Twitter Sentiment Analysis of Independent Learning in Independent Campus Using RoBERTa Base IndoLEM Sentiment Classifier Model

Higher education is one of the fields that is affected by the COVID-19 pandemic. One of the regulations in Indonesian higher education is Merdeka Learning-Independent Campus (MB-KM). In this research article we are aiming to do an analysis of the sentiment using Twitter as the dataset in order to find out the sentiment of university students towards the program. The analysis was being done using RoBERTa Base IndoLEM Sentiment Classifier Model. With the total number of more than 30 thousand positive sentiments and almost 20 thousand negative sentiments found. The result of the model shows that it achieved an accuracy of 91.73%, with a precision of 83.33%, recall 89.74%, and a harmonic mean of the two (F1 score) of 86.42%. Based on the analysis, it also shows the distribution of the sentiment is 63.0% of positive sentiment and 37.0% of negative sentiments. This paper shows that there are more positive sentiments than the negative one. © 2022 IEEE.

Analysing Tweets on COVID-19 Vaccine: A Text Mining Approach

The COVID-19 pandemic has caused large scale health, economic, and social crisis. Scientists throughout the globe have been working on producing effective vaccines to combat this pandemic. COVID-19 vaccine release started in 2020, and low take-up rates among the public have been observed initially. There has been a soar in social media data on vaccines. This paper presents a comprehensive analysis of COVID-19 vaccine-related tweets. Sentiments shared by people through tweets and common topics have been extracted using classification and sentiment analysis. Our results showed a higher negative sentiment when the pandemic was declared, and it gradually changed to positive with the COVID-19 vaccine development/rollout. Tweet sentiment analysis offers health departments around the globe a quick sense of public sentiment towards the vaccine. Dominant topics or areas of concern have been identified using topic modelling that might need to be addressed. © 2022 IEEE.

Sentiment Analysis for Twitter Chatter during the Early Outbreak Period of COVID-19

This study presents the findings of applying sentiment analysis on a corpus of seven million unique English tweets collected from March 26, 2020 to April 9, 2020 about the COVID-19 outbreak. First, an off-the-shelf lexicon-based sentiment analysis tool was used to determine sentiment polarity in each tweet. Then, an off-the-shelf text visualization tool was used to visualize the most frequent emotions and topics that showed positive and negative sentiments. The study revealed meaningful insights about which positive and negative emotion types were most prominent on Twitter chatter during the early period of the COVID-19 pandemic, and which topics garnered the most positive and negative emotional reactions. This work shows that analyzing social media chatter using sentiment analysis and text visualization tools is an effective approach for tracking people's concerns and mental health during pandemics and infectious diseases outbreaks. © 2021 IEEE.

Sentiment Analysis of Indonesian Government Policies in Handling Covid 19 Through Twitter Data

This paper presented a sentiment analysis of the Indonesian government's policies in overcoming Covid 19 through twitter data using several classification methods, namely SVM, Naive Bayes, and LSTM. Based on the analysis of the twitter data, it was found that the twitter community in Indonesia gave negative sentiments to government policies in handling Covid 19. From the experimental results, it was found that SVM gave the best sentiment results compared to Naïve Bayes and LSTM by providing an accuracy of 88.5%. © 2021 IEEE.

The Influence of Social Factors on Mental Health and Wellbeing during the COVID-19 Pandemic

This study uses Natural Language Processing and Machine Learning techniques to understand the effect of the COVID-19 pandemic on mental wellbeing. We considered different user groups and locations in the USA to analyze the influence contrasting social factors, such as political stance, had on wellbeing. We measured the mental wellbeing of the social media users through understanding negative sentiment and mental health topic discussion in Twitter posts added by users from the top 10 Democrat and top 10 Republican cities in the USA. To measure the topic discussion, we used a mental health keyword list and developed machine learning models to classify the topic of a tweet. The primary findings include the similarity of the effect the pandemic had on Republican and Democrat cities when considering a timeline of tweets, whilst an increase in 'Anxiety' was discussed across different user groups and cities. Enforcement strategies had an influence on mental wellbeing with results differing for Republican and Democrat cities. An accurate text classifier was developed and used to categorize tweets into different mental health topics. The results showed how medical and unemployed users discussed topics like 'anxiety' and 'depression' more than a control set of users. The best machine learning model was developed using a Decision Tree algorithm which achieved an accuracy of 87% on unseen data. © 2021 IEEE.

Indonesian Twitter Sentiment Analysis Application on the Covid l9 Vaccine Using Naive Bayes Classifier

It's even one year since the COVID-19 pandemic hit Indonesia, to anticipate it, the government brought in a COVID-19 vaccine. Various types of COVID-19 vaccine have been introduced to Indonesia, including which ones will be considered the best according to the community through the Twitter platform. One of the venues that creates the most public sentiment is Twitter. It can be determined whether the public fully approves or rejects the existence of vaccination in Indonesia by analyzing public sentiment surrounding the COVID-19 vaccine. Data acquisition using a crawling procedure by connecting the Twitter API, pre-processing, sentiment categorization, and sentiment analysis outcomes are the stages of the sentiment analysis process to become a sentiment analysis application. The PHP and MySQL programming languages are used to create the database for the sentiment analysis application. After the application has been fully implemented, it can do sentiment analysis from each dictionary probability using the Naive Bayes Classifier approach. The study of the two keywords "vaksin covid"and "vaksin corona"yielded the following results. It has 93% positive sentiment results, 72% negative sentiment results, and 35% neutral sentiment outcomes, with an accuracy of 94.74% and 75.47% per keyword. Meanwhile, the Sinopharm vaccine, which has the most positive attitude with the terms "vaksin sinovac,""vaksin astrazeneca,""vaksin sinopharm,"and "vaksin nusantara,"has 84 percent tweets with a 74.23% accuracy rate. © 2021 IEEE.

Sentiment Analysis of COVID-19 Vaccine on Twitter Social Media: Systematic Literature Review

This study attempted to review studies on sentiment analysis of COVID-19 vaccine using Twitter as the source data. It was conducted to understand methods to collect, preprocess and classify data by researchers. We used Systematic Literature Review approach to collect, filter and review research papers found. Research papers were collected from IEEE, ScienceDirect, Springer, ACM Digital Library, arXiv, medRxiv and Google Scholar by taking studies published from 2020 until June 2021. First of all, we gathered paper based on title and it resulted forty papers. On the next step, the contents of the papers were examined using exclusion criteria and inclusion criteria to investigate them further. This step filtered out several papers leaving twenty one corresponding papers to be reviewed for this study. We reviewed the remaining papers to answer four predetermined research questions. The frequently used methods to collect data are RTweet, Twint, Twitter API and Tweepy. Several techniques which used by researchers to preprocess Twitter data are stop word removal, remove punctuation and link, case folding, tokenization, stemming, remove duplicate tweet and lemmatization. To classify sentiment tweets, researchers used several machine learning and deep learning methods. BERT as Transformers method was also used by several researchers. Further studies regarding methods and variables or parameters to classify tweet data are still needed. © 2021 IEEE.

A COVID-19 news coverage mood map of Europe

We present a COVID-19 news dashboard which visualizes sentiment in pandemic news coverage in different languages across Europe. The dashboard shows analyses for positive/neutral/negative sentiment and moral sentiment for news articles across countries and languages. First we extract news articles from news-crawl. Then we use a pre-trained multilingual BERT model for sentiment analysis of news article headlines and a dictionary and word vectors -based method for moral sentiment analysis of news articles. The resulting dashboard gives a unified overview of news events on COVID-19 news overall sentiment, and the region and language of publication from the period starting from the beginning of January 2020 to the end of January 2021. © Association for Computational Linguistics