Public Perception Towards Children's COVID-19 Vaccination with Natural Language Processing

doi:10.21203/rs.3.rs-3088966/v1

Download PDF

Research Article

Public Perception Towards Children's COVID-19 Vaccination with Natural Language Processing

https://doi.org/10.21203/rs.3.rs-3088966/v1

This work is licensed under a CC BY 4.0 License

Version 1

posted

You are reading this latest preprint version

In 2019, Coronavirus manifested itself in China and caused numerous deaths. Vaccines developed against COVID-19 are seen as a way to end or mitigate the pandemic. Many debates arose about the vaccination of children through social media. The main target of this study is to present a model that reveals the perception of parents about getting their children vaccinated, extracts the main themes, and determines the emotional changes. With the support of the Octoparse web scraping tool, data was extracted from Twitter when the epidemic turned into a global problem and the discussions about vaccines intensified. Then, using the topic modeling and sentiment analysis techniques under the umbrella of (Natural Language Processing) NLP, main, sub-topics about parents' attitudes were revealed, also vaccine perceptions were detected by performing sentiment analysis. As a result, four topic clusters were determined: “the opinion of the need for the first dose of vaccination according to age”, “the effectiveness of the first dose of vaccine”, “the opinion of the need for vaccination of school-age children”, and “the need for vaccination arising from the protection of unvaccinated children with only mask protection”. With sentiment analysis, it was seen that positive emotions were dominant, and three emotions, namely trust, expectation, and fear, came to the fore. In conclusion, it has been determined that families trust the states and their announcements about getting their children vaccinated, they anticipate new vaccines to be developed, but they are also afraid of the risks that the vaccine will bring to their children.

COVID-19

vaccine

natural language processing (NLP)

topic modeling

sentiment analysis

web scrapping

At the end of 2019, the coronavirus, which originates from the severe acute respiratory syndrome, first emerged in Wuhan, China, spread rapidly to the world, and affected more than 220 countries, upending lives, thus reaching the epidemic status in 2020 and becoming a global threat [1]–[3]. In order to alleviate the epidemic, basic measures were taken by governments, in this way, the public was tried to be protected in the development process of an effective vaccine. However, although the widespread application of the vaccine is required in order to gain herd immunity, the prejudices and uneasiness of the public create obstacles in front of vaccination. While adults were still uneasy about getting themselves vaccinated, with issues such as reopening schools and stretching the measures, the vaccines to be administered to children became a problem, leading to heated debates in the public. It is vital for all of these reasons that policymakers understand public sentiment towards vaccines that will reduce the negative aspects of the epidemic [4].

Children were deprived of their social activities as their lifestyles were affected by situations. Having to keep up with a process out of the order they are accustomed to, they faced changes such as family diseases, social isolation, and online education. As a result, during the epidemic, they began to experience stress, anxiety, and depression [5], [6]. In addition, their personal space has been narrowed due to the intense worry that families have about themselves [7]. In order to deal with these issues, families shared their concerns about the behavioral changes they see in their children during the process of the pandemic situation, the effects of the new system, how to keep their children safe from the infection, and the side effects of their children about the disease, through social media, as the general public does. Recently, social media platforms have become a place that provides information to many users on current issues and allows them to exchange ideas. Online platforms have become an invaluable choice for people to share their perceptions transparently, also, to receive and disseminate information. With the rapid spread of COVID-19, social media has begun to reflect the concerns of society [8], [9]. In this context, Twitter has come to the fore, also, has become a tool for people around the world to share their thoughts and attitudes [10]. The main motivation of the research is that this study can guide research to be carried out based on vaccination in children and policymakers in this process, by using the newest machine learning methods, with a comprehensive model presentation.

This study aims to identify the attitudes of parents about vaccinating their children in the time frame of 1 August 2020-1 November 2021, when the subject is most discussed, by addressing tweets published around the world containing certain keywords in the English language, and to analyze the characteristics of these attitudes. The second part conducts a complete evaluation of the literature. In the third chapter, the methodology was explained in detail, and application steps were shown. In the fourth section, the outputs of the model were discussed. And lastly, in the fifth part, the study was concluded and suggestions about future works were given.

Technology advancements have offered new opportunities for numerous sectors, allowing them to develop and enhance quality, with the healthcare industry benefiting from digital transformation by improving patient care, safety, and capacities [11]. Researchers have turned their focus to COVID-19 and aimed to determine and analyze the echo of the COVID-19 around the world, the public attitude towards the coronavirus vaccine, the perception of the parents about their children in this period, and the attitude of the parents towards a possible vaccine to be given to their children. Thus, they carried out their research by adopting many different data sources, methods, and perspectives. It was established that studies conducted in the early days of the pandemic were focused directly on the epidemic and also discussed policymakers' actions around the world or in a specific country. On contrary, it was observed that the scope of the subject was expanded in the future, and different perspectives were included in the studies conducted.

2.1 COVID-19 echo and vaccine perception in the world

COVID-19 has completely changed the lifestyle of people in different regions and countries; thus, it has become a very important global problem. Coronavirus vaccines have attracted great interest from the public, subsequently, they have led to many discussions that have brought about concerns and hesitations. In this context, the researchers, who emphasized the importance of understanding the repercussion of COVID-19 in the public, have set themselves the goal of examining the troubles experienced by public and parents during the pandemic also, the effects of these troubles on children, as well as analyzing the concerns of families for their children. In this way, they aimed to guide policymakers by examining the problems and concerns for increasing vaccine intention and shed light on the debates about vaccines.

Patrick et al. [12], Zhen [13], Yeasmin et al. [14], Lazarus et al. [15] conducted a research study by using questionnaire to understand parents' perceptions of the consequences for their children and families during the epidemic. Murphy et al. [16], Alfieri et al. [17], Yigit et al. [18], Halvorson et al. [19], Bell et al. [20] carried out a study with the help of surveys, emphasizing that identifying vaccine hesitancy in different samples for children’s and general public.

With the help of online surveys, Ali et al. [21], Fitzpatrick K et al. [22] aimed to analyze the behavior, knowledge, and perceptions of the public regarding the epidemic to identify deficiencies in key parts of public information. Khubchandani et al. [23], Ward et al. [24], Reiter et al. [25], Malik et al. [26], Zhang et al. [27], Skjefte et al. [28], Yılmaz and Sahin [29] took help from social media to get ideas from the general population and conducted online surveys for the vaccine perceptions.

Singh et al. [30], Raamkumar et al. [31], Park et al. [32] aimed to develop a perspective on the pandemic by addressing social media platforms like Twitter, Facebook. Scannell et al. [33], wanted to figure out how different persuasive strategies were employed in COVID-19 vaccination tweets. Also, Jenkins and Moreno [34] aimed to analyze the shares of parents on social media about getting their children vaccinated.

In order to analyze the public's thoughts and discourses about the epidemic, support is often received from online environments. In addition, the traditional method of surveys was carried out from online platforms to find out what the general public thinks, thus, the opinions of the public are taken into consideration transparently. Also, to analyze the perception of vaccines, researchers aimed to determine the existing attitudes by asking specific questions about the vaccine. As a result of surveys or interviews with parents, it was discovered that the COVID-19 showed some results in terms of education and physical/mental health in children. When looking at existing studies, it was shown that relatively few uses major social media platforms other than blogs to profit from machine learning approaches, as a result, the lack of this situation in the literature has been observed. It is thought that conducting a study on the hesitancy of vaccination in children on Twitter, where the public shares their feelings in the most transparent way, will bring clarity to the subject and bring the issue of children and vaccination to the fore with sharp lines throughout the world.

2.2 Natural language processing (NLP) studies in the COVID-19

Natural language processing (NLP) is a field that tries to process, understand, and interpret human language. It provides methods that help transform the text into a structured representation and enables artificial intelligence to make sense of it by taking human language as input. In short, they are techniques that enable automatic extraction and identification of information from texts [35]–[37]. In this context, to analyze the attitudes of the public against the pandemic, researchers have benefited from NLP techniques and examined the main themes and mood changes of public perception within the historical frameworks they determined.

To develop an answering framework to answer medical questions Masum et al. [38] were used the methods of query expansion, data pre-processing and vector space models. Also, Klein et al. [39] was applied the deep neural network classifier based on Bidirectional Encoder Representations from Transformers (BERT) model for Developing a Twitter-based NLP data pipeline. For coping with misinformation during the outbreak, Ayoub et al. [40], were put a framework by getting help from the SHAP and DistilBERT methods. Latent semantic analysis (LSA) was performed for extracting keywords based on demographic, social, epidemiological, economic, psychological, medical, and clinical perspectives of the pandemic with the support of NLP algorithms, based on the literature by Bose et al. [41]. And also, Tang et al. [42] showed the application of BERT for analyzing the tweets about the epidemic posted by health organizations.

Sadman et al. [43] collected news reports about the pandemic and used NLP techniques to obtain information about the number of cases, trending topics, emotions, and the virus. L. Li et al. [10] received support from NLP techniques by processing the data pulled from the Weibo platform with the help of supervised learning to classify the available information about the epidemic into different types of situation information. On the other hand, Samuel et al. [44] aimed to determine the public perception of COVID-19, with the help of sentiment analysis packages of the R program, by analyzing tweets about the pandemic. Osakwe et al. [45] aimed to identify the main issues by using textual data to determine the reaction and concern of the public during the pandemic, with the support of tweets .

Different NLP techniques are used by researchers to evaluate public opinions with the support of social media platforms, to determine the accuracy of information shared by news sources or websites, to determine how much the public trusts the news, information, or ideas that are circulating, and how such articles affect people has been observed. Topic modeling techniques have come to the fore, as researchers who perform unsupervised learning practice mostly use the. Furthermore, it was discovered that the authors made extensive use of sentiment analysis techniques to further strengthen their research and determine the perception of public opinion, by this means, they had the chance to understand, also, interpret the public's feelings about the pandemic.

2.2.1 Topic modeling and sentiment analysis studies on the perception of COVID-19

Topic modeling, also known as probabilistic clustering, is thought to be one of the most effective techniques for classifying and clustering textual data [46], [47]. Researchers prefer to use topic modeling techniques that make use of unsupervised learning to reveal hidden thematic structures from the data sets they have. Also, Sentiment analysis is a computer tool of NLP that researchers use to identify the public's point of view on a topic [48]. With the use of sentiment analysis, people's emotions can be interpreted and the interpreted emotions can be used where necessary for any organization [49], [50]. Techniques related to this aspect have played a key part in the studies carried out to inform policymakers and understand the emotional change-development of the public in the COVID-19 by attracting the attention of researchers.

By using topic modelling, Agarwal et al. [51] developed a system for analyzing tweets about the COVID-19. On the other hand, Doogan et al. [52] identified the posts about non-drug interventions by considering six different countries. Also, Gozzi et al. [46] focused on the public's internet response and media content during the pandemic. Liu et al. [54] compared people's behavior changes, thoughts, confirmed cases, and deaths. Examining media reports about the outbreak and health communication patterns governed by the media Q. Liu et al. [55] determined 9 main themes.

Tsai and Wang [48] aimed to perform a sentiment analysis study with the help of Twitter data, and also, examined the relationship between mood changes in tweets about the pandemic and public health policies. Lwin et al. [56] aimed to analyze the worldwide emotional trends based on four main emotions, namely anger, fear, joy, and sadness, as well as to investigate the underlying causes of the emerging emotions in this direction. Pokharel [57] aimed to conduct a sentiment analysis to determine how people living in different countries coped with the pandemic situation based on 12 selected countries of the tweets sent. According to the tweets shared on Twitter, Dubey [58] aimed to analyze emotions and feelings, as well as to investigate the situation of cyber racism that increased during the epidemic. In order to ascertain the mental state of a people during their health period, Singh et al. [59], who received support from Twitter, performed a sentiment analysis with the help of the BERT model, by performing an application on tweets sent by users. Samuel et al. [44] tried to detect public awareness of the epidemic by using tweets.

Aware of the advantageous aspects of topic modeling and sentiment analysis methods, the researchers thought that using these two techniques could be more efficient in listening to the voice of the public and observing the process. For example, to raise public awareness of the epidemic and revealing issues of concern shared by social media users Boon-Itt and Skunkan [60] were applied these two methods. Yin et al. [61] present a new framework to analyze the emotional dynamics and topics about the epidemic from social media posts. Chen et al. [62] used Twitter to learn about the public's opinion of the pandemic. Hung et al. [63] analyzed and identified sentiments related to the pandemic on Twitter. Also, Abd-Alrazaq et al. [64] defined the main topics shared by Twitter users about the pandemic.

Ebadi et al. [65] aimed to identify hidden issues by examining publication similarities and the change of emotions by using a structural topic modeling framework. Oyebode et al. [66], by using six different social media platforms, aimed to analyze the public's thoughts and comments regarding the COVID-19. Li et al. [67] aimed to examine the tweets of the COVID-19 period in terms of mental health by developing models. After the curfews, Hanschmidt and Kersting [68] aimed to analyze the emotions of the public reflected on social media platforms and to examine the connection with emotions by revealing the main themes discussed. Raju et al. [69] aimed to put forward an NLP model with high classification accuracy based on the BERT model, which will support taking urgent measures to stop the proliferation of the epidemic by cleaning the misinformation on social media for health institutions. Lee [70] aimed to determine the topics around which people's thoughts were shared on Twitter about COVID-19, why these issues came to the fore, and how emotions developed during the epidemic. Vydra and Kantorowicz [71] aimed to determine how useful social media data are for policymakers using latent semantic scaling and structural topic models. Focusing on different periods, M. S. Ahmed et al. [72] aimed to examine the sentiment of users and support trending topics using a dataset of tweets about the epidemic. Fazeli et al. [73] conducted a study when the epidemic first emerged, aiming to present a framework and present a multi-faceted analysis of the critical features reflected by the conversations on social media.

It was observed that the researchers, by pulling the main themes of the epidemic in news, articles, or tweets, divided the perceptions into different clusters, and analyzed what perspectives were formed in the public. According to the research, Latent Dirichlet Allocation (LDA), which is one of the most effective algorithms in technically unsupervised machine learning applications, is frequently utilized as a data source to obtain public information. Researchers, focusing on understanding the psychological impact of the pandemic on people, thought that the best way for this was sentiment analysis, and focused on analyzing perceptions related to the epidemic by making use of different methods. With the help of two NLP techniques, more comprehensive results were revealed, and both main themes and emotions were determined with sharp lines. At this point, the striking point is that social media platforms are predominantly preferred in studies, and it has been determined that Twitter is the most used social media platform to obtain clearer data. As a consequence, it has been observed that these two techniques are used separately in the articles, to the best of our knowledge, there hasn't been any research that combines these two methods and puts forward a model.

2.2.2 COVID-19 vaccine perception with topic modeling and sentiment analysis

Researchers aimed to reveal the attitudes of the public toward COVID-19 vaccines, which are thought to be a way to prevent COVID-19. In addition, they aimed to present studies that approached the event thematically and determined the views of the society to understand people's perspectives on both a possible vaccine and the current vaccine, in this way, they benefited from topic modeling techniques. Managers and health policymakers used evidence-based scientific data to guide decision-making [74], [75] [76]. Focusing on analyzing and classifying text emotions, sentiment analysis can describe people's perspectives on public health policies for policymakers [48]. In this way, subjective opinions can be categorized within themselves and gain meaning. With this respect, sentiment analysis techniques attract the attention of the authors and have been widely used in examining the public acceptance of vaccines produced or being developed for the coronavirus. The authors, who want to take the advantageous aspects of the topic modeling and sentiment analysis methods under the umbrella of NLP, received support from these techniques to determine the attitudes about the recently popular COVID-19 vaccine. In this way, they have introduced models to the literature that extract key themes and sentiments about vaccines.

To uncover themes found in the public's thinking about possible vaccines early in the COVID-19 Jiang et al. [77], to observe temporal changes by uncovering themes in tweets S. Liu et al. [54], to analyse comments on social media about parents, children, vaccines, and COVID-19 Thurson [78] used LDA method. Also by applying sentiment analysis, Ahmed et al. [49] analyzed public opinion about the COVID-19 vaccine, Sattar and Arifuzzaman [79] determined vaccine insights and examining the public's perspective on safety measures after intake and Hussain et al. [4] developed an artificial intelligence-based model. Gbashi et al. [80] performed systematic review for vaccine opinion.

To be able to identify the main themes and feelings discussed the vaccine Lyu et al. [81] used topic modelling and sentiment analaysis methods together. Additionally, to deduct topics and feelings about the vaccine Kwok et al. [82] put forward a model. Shim et al. [1] analyzed, interpreting, and classifying tweets about the vaccine. Karami et al. [83] determined the sentiment of tweets, pinpointing key issues, examining change, and revealing key points.

In topic modeling studies, it has been observed that Twitter is generally used to pay attention to the voice of the public by using the LDA method, but besides Twitter, blogs, articles, and news sites are examined. The authors identified the main themes, made recommendations to policymakers, and examined the public's thoughts, confidence, and discourses against vaccination. In order to analyze the mood of the people against the vaccines to be made to prevent the epidemic, the authors benefited from different sentiment analysis techniques. It has been emphasized that states and policymakers should be more transparent and clearer to the public about the vaccine to prevent these concerns.

While the authors aimed to identify the main topics and feelings from the ideas, and discourses of the public using vaccine-related keywords, they preferred social media, especially Twitter. It has been observed that researchers mostly benefit from LDA, which is one of the unsupervised learning techniques of topic modelling. It was found that control measures against the vaccine were generally supported, and false information was rejected. Conspiracy theories have created a shield against the vaccination even though it has been noticed that positive emotions are not enough to reach the amount of vaccine needed to achieve herd immunity. It was emphasized to governments that they should pay attention to the people's voices, and also, be pioneers in supporting vaccine development and clinical management by creating a vaccine promotion plan.

The fundamental purpose of this research is to look into the perspectives of parents’ attitudes and beliefs regarding getting their children vaccinated against coronavirus, which is critical for herd immunity, and to understand the reasons underlying these beliefs. In this regard, it is targeted to analyze the data set, which was taken from Twitter within certain limits, by applying topic modeling and sentiment analysis methods, with the support of the R software program (see Fig. 1).

3.1 Web scraping and dataset generation

Web scraping is the process of gathering unstructured data from the internet and combining it with other data to create large-scale structured datasets [84]. Twitter is a social media platform where people freely share their views, bring together people from different regions, and have millions of information on almost every subject. For this reason, it is a platform preferred by researchers to create large data sets in the process of performing analysis on a specific topic. Based on the literature analysis, Twitter was chosen as the online venue for the construction of the data set in this study, and 59,928 tweets indicating families' perspectives on the coronavirus vaccine to be administered to their children were captured using the Octoparse software tool.

Octoparse is a piece of software that allows you to scrape any desired website offering a free plan up to a certain point to generate datasets in different formats. It is an easy-to-use yet powerful tool used to create the dataset [85]. Due to all the advantages, it provides, it was decided to use it to extract data from Twitter in this study, and it provided great benefits with the ease of use during the creation of the data set. In this direction, data was extracted from Twitter with the help of the Octoparse web scraping tool, using keywords determined in English in order to address the entire world, between August 1, 2020, and October 1, 2021, when the epidemic became a global problem and vaccine discussions became more intense.

Firstly, as a consequence of the literature review, the keywords to be searched on Twitter, which can pull tweets based on specified topics, were determined (see Table 1).

Vaccine related keywords

COVID related keywords

Child-related keywords

• Vaccine

• Vaccination

• Biontech

• Sinovac

• Vaccinal

• Immunization

• Covid

• Corona

• Coronavirus

• Covid19

• Covid-19

• Pandemic

• Epidemic

• Children

• Child

• Kid

• Kids

• Boy

• Daughter

• Boys

• Daughters

• Youngster

• Junior

• Tiddler

• Son

• Parents

3.2 Pre-processing

R is a programming language and software environment for reporting, graphical representation, and statistical analysis [86]. It has been preferred for the analysis to be used in this study, as it provides different visual outputs, facilitates the interpretation of the outputs, and provides rapid processing of a large amount of data at the same time. The data set created for the preprocessing process was transferred to R Studio and the preprocessing process was started. As a result, the information was cleaned and made ready for analysis. The cleaning procedure was carried out with the assistance of the functions, involving removing duplications, converting characters to lower case, removing URLs, removing usernames, removing hashtags, removing carriage, removing emoticons, removing numbers, removing punctuation, removing anything except the English language, and space, removing other invoices, removing stop words, eliminating terms that were utilized as a keyword in the Twitter search, eliminating extra whitespace, lemmatization and creating document-term frequency matrix.

The preprocessing of the input is of great importance in analyzes using NLP methods. This is the process of cleaning and preparing the input texts for analysis, and it is considered the initial stage in text classification. Data gathered from social media platforms, in particular, contains a lot of noise. There is a lot of data in datasets made using Twitter that does not contain meaningful information for analysis. By performing the preprocessing before the analysis, these data, which are called noise and have no benefit to the analysis, are removed from the data set [87].

3.3 Topic modelling

Topic modeling is an unsupervised machine learning method frequently encountered in sciences such as NLP, information retrieval, and text mining, which creates a statistical model and is used to find the number of words frequently used in textual information [88]. The methods under this heading assume that meanings are composed of word sets that describe related topics of conversation while at the same time accepting meanings as relational [89]. Using this technique, decision-makers observe frequently used word patterns in data sets to obtain meaning, thus revealing the data's hidden information, that is, the topic of the data, and labeling them accordingly (see Fig. 2). Through analysis, researchers can discover hidden themes in the collection, group documents into discovered themes, and summarize documents by topic [65]. Unsupervised topic models can summarize, understand, analyze, and organize information in large data sets.

LDA is a generative statistical model that explains why some parts of a dataset are similar to each other but is the most popular topic modeling and analysis term, which assumes that there are certain topics known as fixed word distributions for the entire document collection [91]. It works by assuming that the topics make up the document and that each topic creates the words that make up the document, based on the probability distribution [92], [93]. LDA is widely regarded as a useful method for uncovering and exploiting a huge dataset's hidden thematic structure [94].

The distribution used to determine the distribution of topics per document is known as the Dirichlet distribution. Then, in the generative process for the LDA method, the results from this distribution are used to allocate words in the document to different topics. While documents are observable in LDA, topics, distribution of topics per document, and classification of each word by topics per document are hidden constructs. This is where the algorithm got its name [95].

In Fig. 3, the graphical model representation of LDA was given. Replicas are represented by the boxes, which are "plates”. The outer plate represents documents, while the inner plate reflects the document's repeated selection of topics and words. In Fig. 3 [96];

α gives the distribution of topics per document,

β gives the word distribution per topic,

Θ indicates the distribution of topics for a particular document,

z is the assigned topics for each word,

w are observed words.

The α and β parameters are sampled once during system creation. The Θ parameter is sampled for each document in the system. Also, the logic of the method can be supported by Fig. 4 for better presentation.

The LDA algorithm, as mentioned before, is based on a statistical basis. However, a normal statistical model assumes that for each of the vocabulary words, there is a probability of that word occurring. LDA adds a layer of complexity to this system. LDA counts a list of topics (k). Each document (m) is a probability distribution over k topics. Each topic is a probability distribution over all the terms in our vocabulary (V). It accepts that every word has various possibilities of appearing in every topic. The exact probability formula that LDA uses and creates a document is explained in the following [98];

The first two sums in the formula contain symmetrical Dirichlet distributions, which are called preliminary probability distributions for topics and documents. The third total includes two polynomial distributions, one on words and one on topics. By using the final probability formula, the same word distribution obtained in the original documents is tried to be created [98].

3.4 Sentiment Analysis

Sentiment analysis techniques aim to reveal the emotions that exist for any topic in a text. Thus, it generally categorizes texts as positive, negative, or neutral based on emotion scores [99]. In short, it's the process of quickly reporting and making sense of massive amounts of data using software systems, with burgeoning social networks at the forefront. With the feature of sentiment analysis to question whether the texts have positive, negative, or neutral content in general, the attitude of the individuals concerned about the relevant topic is determined [100]. Sentiment analysis is a natural language processing approach used to measure an idea or emotion expressed in a series of tweets, which was applied in this study (see Fig. 5) [101].

In this study, a model that provides the general attitude of families against the coronavirus vaccine in children was presented by analysing tweets that contain the ideas of parents against the vaccine for their children. While establishing the model, analysis methods were used under the umbrella of NLP. Therefore, the conclusion part can be considered as two stages (see Fig. 6). The first is the establishment of topic clusters by applying the LDA method, one of the topic modelling techniques that reveal the most talked about topics. The second one is the stage of performing sentiment analysis, which reveals attitudes by labelling the generated topic clusters to the data.

LDA method, which is one of the topic modeling methods, was used to reveal the topic clusters and to observe the most frequently mentioned words. In this context, with the help of the web scraping tool, 59,928 tweets were retrieved as a result of an advanced search on Twitter. First, it was discovered that when the terms were organized according to their frequency in the data set (see Fig. 7), the clearest question mark that arose in the thoughts of the families regarding the vaccine was related to their children's school status.

4 topic clusters and the 5 most spoken terms for each topic cluster are presented in Fig. 8. Then, topic clusters were tagged in tweets and saved as a CSV file for later use in sentiment analysis.

Figure 8 was created according to beta values based on the probability of words being found in documents to improve vision and execute interpretations. When the results are analyzed one by one on a cluster basis, the main themes revealed by the topic clusters with the terms that make them up were continuously formed and interpreted;

The opinion of the need for first dose vaccination by age

First dose vaccine efficacy

The idea of the need for vaccination of school-aged children

The need for vaccinations arising from the protection of unvaccinated children only with a mask protection.

With the determination of these 4 topics, sentiment analysis methods were used to determine the attitudes of families in getting their children vaccinated in more detail and to put forward a model. Thus, with the help of the topics tagged to each tweet in the dataset, the sentiment score was calculated for each topic cluster. In line with the evaluations made, the information about which words the prominent emotions are gathered around is of great importance for the analysis.

An investigation of the tweets referring to the first cluster of topics found a connection between the concept of vaccinating families for age-related reasons and the faith placed in governmental organizations after completing sentiment analysis using existing methodologies (see Fig. 9). Furthermore, a widespread feeling of expectancy was detected. Observations revealed that families were willing to vaccinate their children in this setting and trusted the suggestions of professionals such as doctors, nurses, and teachers who had been supportive throughout the pandemic. While the findings mostly represented positive emotions, a significant fear of the risks and potential adverse effects of vaccination was observed among families in the initial subset of patients.

As illustrated in Fig. 10, families had a typically positive view of the efficacy of the first immunization dose. Trust, anticipation, and dread are prominently conveyed, as were the observations in the first topic cluster. The data imply that faith in the vaccination's effectiveness hinges around the COVID-19 vaccine, which was developed and released globally by presidents. Furthermore, there is an increased expectation for the creation of new vaccinations to obtain more potent results. Despite this prevalent confidence, it has been observed that when the debate focuses on their children, families express anxiety and apprehension about potential threats.

When evaluating families' thoughts on the importance of vaccinating their school-age children, a strong presence of positive emotions and a strong sense of trust appeared (see Fig. 11). It has been seen that families are eager to resume in-person schooling if a suitable climate is created, specifically by requiring required immunization for all school workers and pupils. Furthermore, parents have shown a willingness to investigate the development of innovative vaccinations and provide the first dosage to their children, but with reservations about potential hazards. As a result, parents see school staff immunization as a way to protect their unvaccinated children.

Positive emotions were shown to be more prevalent in the fourth topic cluster (see Fig. 12), although closely associated with negative attitudes. The dominant feelings were trust, fear, and anticipation. It has been determined that families believe in the efficacy of wearing masks as a preventative strategy while also anticipating the arrival of a new vaccination. Concerning their children, however, it has become clear that parents are concerned about potential infectious conditions and the probable implications of vaccination.

As a result, it was seen that positive emotions predominated in the 4 topic clusters and three emotions came to the fore;

Trust
Anticipation
Fear

It is thought that the most important reason for this is that children are at a very sensitive point for their families. Families respect the strategic measures taken by the government or decision-makers and trust their protection. But with the relaxation of the measures taken, they are worried about their children who have no immunity to the coronavirus. This brings the vaccine debate to the fore. With the return to normal social life and the opening of schools, families have come to expect the coronavirus vaccines to be given to their children. In this way, they follow the developments all over the world, also, they are anticipation to develop existing vaccines and conducting new vaccine trials. However, when it comes to their children, they are afraid of the risks that the vaccine will bring, and they are afraid of the side effects of this coronavirus vaccine.

At the end of 2019, the coronavirus epidemic emerged in Wuhan and became a common problem for everyone. Society accepted that the most important way to cope with COVID-19 is the vaccines developed against this disease. While the discussions on vaccines continue, families have created a big question mark in the minds of how the vaccination process of children will proceed. The reactions of families have echoed through popular social media platforms. In this study, it is aimed to analyze the attitudes of families about getting their children vaccinated and determine the reasons behind these attitudes. To the best of our knowledge, no study has been found in the literature that reveals parents' attitudes toward vaccination of their children using a single model that combines topic modeling and sentiment analysis techniques.

Twitter social media platform was preferred in this study, and the related data were created as links by using an advanced search with the support of keywords. In order to make this data workable, the Octoparse tool, which is frequently used by researchers and has a simple structure, was used. Thus, a data set containing tweets of families about children’s vaccination was created. With the help of the created dataset and RStudio, the targeted model has been put forward. First, data preprocessing was performed. Then, topic clusters were created with the support of LDA, and the created clusters were processed with sentiment analysis, thus, the views of families on getting their children vaccinated were revealed with the support of the established model.

As a result of the study, four topic clusters were determined which were the opinion of the need for first dose vaccination by age, first dose vaccine efficacy, the idea of the need for vaccination of school-aged children, and the need for vaccinations arising from the protection of unvaccinated children only with mask protection according to frequently used words. By the processing of the determined topic clusters with sentiment analysis, it was determined that positive emotions were predominant, and 3 emotions, namely trust, expectation, and fear, came to the fore.

During the study period, some limitations occurred in the data collection phase. Since this study cannot focus on each language separately, it has been focused only on tweets sent in English, which is a universal language, and this limited the target audience. In addition, the sentiment libraries of RStudio were used in the sentiment analysis phase, and emotion scores were calculated. In future studies, it is thought that libraries can be created according to the data set to be used in the sentiment analysis phase to achieve clearer and desired results with higher accuracy.

The increasing diversity and importance of NLP studies in recent years to catch up with the process of the coronavirus have been demonstrated by a detailed study. It is thought that the study will shed light on the professionals who develop strategies in the field of health during the process development stages and the researchers in their study.

Author Contributions The corresponding author is Elif Delice (E-Mail: [email protected]). Contributing authors E.D.: Investigation, methodology, software, writing and original draf. H.T.: Supervision, conceptualization, formal analysis, writing and editing. O.P.: Audit, validation review. M.K.: Validation and editing.

Funding No funds or financial support were received during the preparation of the manuscript

Conflict of interest The authors declare no conflict of interest.

Data availability The data that support the findings of this study are available from the corresponding author, ED, upon reasonable request.

Competing Interests The authors have no competing interests to declare that are relevant to the content of this article and there are no financial interests

J. G. Shim, K. H. Ryu, S. H. Lee, E. A. Cho, Y. J. Lee, and J. H. Ahn, “Text mining approaches to analyze public sentiment changes regarding covid-19 vaccines on social media in korea,” Int. J. Environ. Res. Public Health, vol. 18, no. 12, 2021, doi: 10.3390/ijerph18126549.
K. A. Fisher, S. J. Bloomstone, J. Walder, S. Crawford, H. Fouayzi, and K. M. Mazor, “Attitudes Toward a Potential SARS-CoV-2 Vaccine : A Survey of U.S. Adults,” Ann. Intern. Med., vol. 173, no. 12, pp. 964–973, 2020, doi: 10.7326/M20-3569.
“COVID Live Update: 244,536,555 Cases and 4,966,160 Deaths from the Coronavirus - Worldometer.” .
A. Hussain et al., “Artificial intelligence-enabled analysis of public attitudes on facebook and twitter toward COVID-19 vaccines in the United Kingdom and the United States: Observational study,” J. Med. Internet Res., vol. 23, no. 4, pp. 1–10, 2021, doi: 10.2196/26627.
P. G. Szilagyi et al., “Parents’ Intentions and Perceptions About COVID-19 Vaccination for Their Children: Results From a National Survey,” Pediatrics, vol. 148, no. 4, p. e2021052335, 2021, doi: 10.1542/peds.2021-052335.
A. Chandarakesan, S. Muruhan, and R. R. A. Sayanam, “Morin Inhibiting Photocarcinogenesis by Targeting Ultraviolet-B-Induced Oxidative Stress and Inflammatory Cytokines Expression in Swiss Albino Mice,” Int. J. Nutr. Pharmacol. Neurol. Dis. |, vol. 8, pp. 41–46, 2018, doi: 10.4103/ijnpnd.ijnpnd.
M. Spinelli, F. Lionetti, M. Pastore, and M. Fasolo, “Parents’ Stress and Children’s Psychological Problems in Families Facing the COVID-19 Outbreak in Italy,” Front. Psychol., vol. 11, no. January, pp. 1–7, 2020, doi: 10.3389/fpsyg.2020.01713.
M. Drouin, B. T. McDaniel, J. Pater, and T. Toscos, “How Parents and Their Children Used Social Media and Technology at the Beginning of the COVID-19 Pandemic and Associations with Anxiety,” Cyberpsychology, Behav. Soc. Netw., vol. 23, no. 11, pp. 727–736, 2020, doi: 10.1089/cyber.2020.0284.
“How Many People Use Social Media in 2021? (65+ Statistics).” .
L. Li et al., “Characterizing the Propagation of Situational Information in Social Media during COVID-19 Epidemic: A Case Study on Weibo,” IEEE Trans. Comput. Soc. Syst., vol. 7, no. 2, pp. 556–562, 2020, doi: 10.1109/TCSS.2020.2980007.
E. Delice, L. Ö. Polatlı, K. A. Jbara, H. Tozan, and A. Ertürk, “Digitalization in Healthcare : A Systematic Review of the Literature †,” no. 1, pp. 10–11, 2023.
S. W. Patrick et al., “Well-being of Parents and Children During the COVID-19 Pandemic: A National Survey,” Pediatrics, vol. 146, no. 4, p. e2020016824, 2020, doi: 10.1542/peds.2020-016824.
W. Zhen, “Parental Perception about the Effects of Covid-19 Pandemic on Young Children By,” 2020.
S. Yeasmin et al., “Impact of COVID-19 pandemic on the mental health of children in Bangladesh: A cross-sectional study,” Child. Youth Serv. Rev., vol. 117, no. July, p. 105277, 2020, doi: 10.1016/j.childyouth.2020.105277.
J. V. Lazarus et al., “A global survey of potential acceptance of a COVID-19 vaccine,” Nat. Med., vol. 27, no. 2, pp. 225–228, 2021, doi: 10.1038/s41591-020-1124-9.
J. Murphy et al., “Psychological characteristics associated with COVID-19 vaccine hesitancy and resistance in Ireland and the United Kingdom,” Nat. Commun. 2021 121, vol. 12, no. 1, pp. 1–15, Jan. 2021, doi: 10.1038/s41467-020-20226-9.
N. L. Alfieri et al., “Parental COVID-19 vaccine hesitancy for children : vulnerability in an urban hotspot,” pp. 1–9, 2021.
M. Yigit, A. Ozkaya-Parlakay, and E. Senel, “Evaluation of COVID-19 Vaccine Refusal in Parents,” Pediatr. Infect. Dis. J., vol. 40, no. 4, pp. E134–E136, 2021, doi: 10.1097/INF.0000000000003042.
M. M. Davis, J. S. Zickafoose, A. E. Halvorson, and S. Patrick, “Parents' Likelihood to Vaccinate Their Children and Themselves Against COVID-19,” medRxiv, p. 2020.11.10.20228759, 2020.
S. Bell, R. Clarke, S. Mounier-Jack, J. L. Walker, and P. Paterson, “Parents’ and guardians’ views on the acceptability of a future COVID-19 vaccine: A multi-methods study in England,” Vaccine, vol. 38, no. 49, pp. 7789–7798, 2020, doi: 10.1016/j.vaccine.2020.10.027.
K. F. Ali, S. Whitebridge, M. H. Jamal, M. Alsafy, and S. L. Atkin, “Perceptions, knowledge, and behaviors related to COVID-19 among social media users: Cross-sectional study,” J. Med. Internet Res., vol. 22, no. 9, pp. 1–10, 2020, doi: 10.2196/19913.
Fitzpatrick K, Harris C, and Drawve G, “Fear of COVID-19 and the Mental Health Consequences in America. Psychological Trauma: Theory, Research, Practice, and Policy,” vol. 12, pp. 17–21, 2020.
J. Khubchandani, S. Sharma, J. H. Price, M. J. Wiblishauser, M. Sharma, and F. J. Webb, “COVID-19 Vaccination Hesitancy in the United States: A Rapid National Assessment,” J. Community Health, vol. 46, no. 2, pp. 270–277, 2021, doi: 10.1007/s10900-020-00958-x.
J. K. Ward et al., “The French public’s attitudes to a future COVID-19 vaccine: The politicization of a public health issue,” Soc. Sci. Med., vol. 265, pp. 2016–2021, 2020, doi: 10.1016/j.socscimed.2020.113414.
P. L. Reiter, M. L. Pennell, and M. L. Katz, “Since January 2020 Elsevier has created a COVID-19 resource centre with free information in English and Mandarin on the novel coronavirus COVID- 19 . The COVID-19 resource centre is hosted on Elsevier Connect , the company ’ s public news and information,” Vaccine, vol. 38, no. January, pp. 6500–6507, 2020.
A. A. Malik, S. A. M. McFadden, J. Elharake, and S. B. Omer, “Determinants of COVID-19 vaccine acceptance in the US,” EClinicalMedicine, vol. 26, p. 100495, 2020, doi: 10.1016/j.eclinm.2020.100495.
K. C. Zhang et al., “Parental acceptability of COVID-19 vaccination for children under the age of 18 years: Cross-sectional online survey,” JMIR Pediatr. Parent., vol. 3, no. 2, pp. 1–13, 2020, doi: 10.2196/24827.
M. Skjefte et al., “COVID-19 vaccine acceptance among pregnant women and mothers of young children: results of a survey in 16 countries,” Eur. J. Epidemiol., vol. 36, no. 2, pp. 197–211, 2021, doi: 10.1007/s10654-021-00728-6.
M. Yılmaz and M. K. Sahin, “Parents’ willingness and attitudes concerning the COVID-19 vaccine: A cross-sectional study,” Int. J. Clin. Pract., vol. 75, no. 9, pp. 1–11, 2021, doi: 10.1111/ijcp.14364.
L. Singh et al., “A first look at COVID-19 information and misinformation sharing on Twitter.,” ArXiv, 2020.
A. S. Raamkumar, S. G. Tan, and H. L. Wee, “Measuring the Outreach Efforts of Public Health Authorities and the Public Response on Facebook during the COVID-19 Pandemic in Early 2020: Cross-Country Comparison,” J. Med. Internet Res., vol. 22, no. 5, 2020, doi: 10.2196/19334.
H. W. Park, S. Park, and M. Chong, “Conversations and medical news frames on twitter: Infodemiological study on COVID-19 in South Korea,” J. Med. Internet Res., vol. 22, no. 5, 2020, doi: 10.2196/18897.
D. Scannell et al., “COVID-19 Vaccine Discourse on Twitter: A Content Analysis of Persuasion Techniques, Sentiment and Mis/Disinformation,” J. Health Commun., vol. 26, no. 7, pp. 443–459, 2021, doi: 10.1080/10810730.2021.1955050.
M. C. Jenkins and M. A. Moreno, “Vaccination Discussion among Parents on Social Media: A Content Analysis of Comments on Parenting Blogs,” J. Health Commun., vol. 25, no. 3, pp. 232–242, 2020, doi: 10.1080/10810730.2020.1737761.
I. J. B. Young, S. Luz, and N. Lone, “A systematic review of natural language processing for classification tasks in the field of incident reporting and adverse event analysis,” Int. J. Med. Inform., vol. 132, p. 103971, 2019, doi: 10.1016/j.ijmedinf.2019.103971.
E. Pons, L. M. M. Braun, M. G. M. Hunink, and J. A. Kors, “Natural language processing in radiology: A systematic review,” Radiology, vol. 279, no. 2, pp. 329–343, 2016, doi: 10.1148/radiol.16142770.
M. A. Zakkar and D. J. Lizotte, “Analyzing Patient Stories on Social Media Using Text Analytics,” J. Healthc. Informatics Res., vol. 5, no. 4, pp. 382–400, 2021, doi: 10.1007/s41666-021-00097-5.
M. Masum et al., “Actionable Knowledge Extraction Framework for COVID-19,” Proc. - 2020 IEEE Int. Conf. Big Data, Big Data 2020, pp. 4036–4041, 2020, doi: 10.1109/BigData50022.2020.9378398.
A. Z. Klein, A. Magge, K. O’Connor, J. I. F. Amaro, D. Weissenbacher, and G. G. Hernandez, “Toward Using Twitter for Tracking COVID-19:A Natural Language Processing Pipeline and Exploratory Data Set,” J. Med. Internet Res., vol. 23, no. 1, 2021, doi: 10.2196/25314.
J. Ayoub, X. J. Yang, and F. Zhou, “Combat COVID-19 infodemic using explainable natural language processing models,” Inf. Process. Manag., vol. 58, no. 4, p. 102569, 2021, doi: 10.1016/j.ipm.2021.102569.
P. Bose, S. Roy, and P. Ghosh, “A Comparative NLP-Based Study on the Current Trends and Future Directions in COVID-19 Research,” IEEE Access, vol. 9, pp. 78341–78355, 2021, doi: 10.1109/ACCESS.2021.3082108.
L. Tang et al., “Texas public agencies’ tweets and public engagement during the COVID-19 pandemic: Natural language processing approach,” JMIR Public Heal. Surveill., vol. 7, no. 4, 2021, doi: 10.2196/26720.
N. Sadman, N. Anjum, K. Datta Gupta, and M. A. Parvez Mahmud, “Understanding the Pandemic through Mining Covid News Using Natural Language Processing,” 2021 IEEE 11th Annu. Comput. Commun. Work. Conf. CCWC 2021, pp. 362–367, 2021, doi: 10.1109/CCWC51732.2021.9376002.
J. Samuel, G. G. M. N. Ali, M. M. Rahman, E. Esawi, and Y. Samuel, “COVID-19 public sentiment insights and machine learning for tweets classification,” Inf., vol. 11, no. 6, pp. 1–22, 2020, doi: 10.3390/info11060314.
Z. T. Osakwe, I. Ikhapoh, B. K. Arora, and O. M. Bubu, “Identifying public concerns and reactions during the COVID-19 pandemic on Twitter: A text-mining analysis,” Public Health Nurs., vol. 38, no. 2, pp. 145–151, 2021, doi: 10.1111/phn.12843.
N. Gozzi et al., “Collective Response to Media Coverage of the COVID-19 Pandemic on Reddit and Wikipedia: Mixed-Methods Analysis,” J. Med. Internet Res., vol. 22, no. 10, pp. 1–19, 2020, doi: 10.2196/21597.
A. Al-Rawi, K. Grepin, X. Li, R. Morgan, C. Wenham, and J. Smith, “Investigating Public Discourses Around Gender and COVID-19: a Social Media Analysis of Twitter Data,” J. Healthc. Informatics Res., vol. 5, no. 3, pp. 249–269, 2021, doi: 10.1007/s41666-021-00102-x.
M. H. Tsai and Y. Wang, “Analyzing twitter data to evaluate people’s attitudes towards public health policies and events in the era of covid-19,” Int. J. Environ. Res. Public Health, vol. 18, no. 12, 2021, doi: 10.3390/ijerph18126272.
A. Ahmed, C. Argho, D. Suben, and K. Saha, “Sentiment Analysis of COVID-19 Vaccination from Survey Responses in Bangladesh,” pp. 1–15, 2021.
C. H. Chang, M. Monselise, and C. C. Yang, “What Are People Concerned About During the Pandemic? Detecting Evolving Topics about COVID-19 from Twitter,” J. Healthc. Informatics Res., vol. 5, no. 1, pp. 70–97, 2021, doi: 10.1007/s41666-020-00083-3.
A. Agarwal, P. Salehundam, S. Padhee, W. L. Romine, and T. Banerjee, “Leveraging Natural Language Processing to Mine Issues on Twitter during the COVID-19 Pandemic,” Proc. - 2020 IEEE Int. Conf. Big Data, Big Data 2020, pp. 886–891, 2020, doi: 10.1109/BigData50022.2020.9378028.
C. Doogan, W. Buntine, H. Linger, and S. Brunt, “Public perceptions and attitudes toward covid-19 nonpharmaceutical interventions across six countries: A topic modeling analysis of twitter data,” J. Med. Internet Res., vol. 22, no. 9, pp. 1–11, 2020, doi: 10.2196/21419.
Y. Liu, C. Whitfield, T. Zhang, A. Hauser, T. Reynolds, and M. Anwar, “Monitoring COVID-19 pandemic through the lens of social media using natural language processing and machine learning,” Heal. Inf. Sci. Syst., vol. 9, no. 1, pp. 1–16, 2021, doi: 10.1007/s13755-021-00158-4.
S. Liu, J. Li, and J. Liu, “Leveraging transfer learning to analyze opinions, attitudes, and behavioral intentions toward COVID-19 vaccines: Social media content and temporal analysis,” J. Med. Internet Res., vol. 23, no. 8, 2021, doi: 10.2196/30251.
Q. Liu et al., “Health communication through news media during the early stage of the covid-19 outbreak in China: Digital topic modeling approach,” J. Med. Internet Res., vol. 22, no. 4, 2020, doi: 10.2196/19118.
M. O. Lwin et al., “Global sentiments surrounding the COVID-19 pandemic on Twitter: Analysis of Twitter trends,” JMIR Public Heal. Surveill., vol. 6, no. 2, pp. 1–4, 2020, doi: 10.2196/19447.
B. P. Pokharel, “Twitter Sentiment Analysis During Covid-19 Outbreak in Nepal,” SSRN Electron. J., no. March, pp. 1–9, 2020, doi: 10.2139/ssrn.3624719.
A. D. Dubey, “The resurgence of cyber racism during the COVID-19 pandemic and its aftereffects: Analysis of sentiments and emotions in tweets,” JMIR Public Heal. Surveill., vol. 6, no. 4, pp. 1–7, 2020, doi: 10.2196/19833.
M. Singh, A. K. Jakhar, and S. Pandey, “Sentiment analysis on the impact of coronavirus in social life using the BERT model,” Soc. Netw. Anal. Min., vol. 11, no. 1, pp. 1–11, 2021, doi: 10.1007/s13278-021-00737-z.
S. Boon-Itt and Y. Skunkan, “Public perception of the COVID-19 pandemic on twitter: Sentiment analysis and topic modeling study,” JMIR Public Heal. Surveill., vol. 6, no. 4, pp. 1–17, 2020, doi: 10.2196/21978.
H. Yin, S. Yang, and J. Li, “Detecting Topic and Sentiment Dynamics Due to COVID-19 Pandemic Using Social Media,” Lect. Notes Comput. Sci. (including Subser. Lect. Notes Artif. Intell. Lect. Notes Bioinformatics), vol. 12447 LNAI, no. June, pp. 610–623, 2020, doi: 10.1007/978-3-030-65390-3_46.
L. Chen, H. Lyu, T. Yang, Y. Wang, and J. Luo, “In the Eyes of the Beholder: Analyzing Social Media Use of Neutral and Controversial Terms for COVID-19,” pp. 3–6, 2020.
M. Hung et al., “Social network analysis of COVID-19 sentiments: Application of artificial intelligence,” J. Med. Internet Res., vol. 22, no. 8, pp. 1–13, 2020, doi: 10.2196/22590.
A. Abd-Alrazaq, D. Alhuwail, M. Househ, M. Hai, and Z. Shah, “Top concerns of tweeters during the COVID-19 pandemic: A surveillance study,” J. Med. Internet Res., vol. 22, no. 4, pp. 1–9, 2020, doi: 10.2196/19016.
A. Ebadi, P. Xi, S. Tremblay, B. Spencer, R. Pall, and A. Wong, “Understanding the temporal evolution of COVID-19 research through machine learning and natural language processing,” Scientometrics, vol. 126, no. 1, pp. 725–739, 2021, doi: 10.1007/s11192-020-03744-7.
O. Oyebode et al., “COVID-19 Pandemic: Identifying Key Issues using Social Media and Natural Language Processing,” 2020, [Online]. Available: http://arxiv.org/abs/2008.10022.
I. Li, Y. Li, T. Li, S. Alvarez-Napagao, D. Garcia-Gasulla, and T. Suzumura, “What are we depressed about when we talk about COVID-19: Mental health analysis on tweets using natural language processing,” Lect. Notes Comput. Sci. (including Subser. Lect. Notes Artif. Intell. Lect. Notes Bioinformatics), vol. 12498 LNAI, pp. 358–370, 2020, doi: 10.1007/978-3-030-63799-6_27.
F. Hanschmidt and A. Kersting, “Emotions in Covid-19 Twitter discourse following the introduction of social contact restrictions in Central Europe,” J. Public Heal., 2021, doi: 10.1007/s10389-021-01613-y.
R. Raju, S. Bhandari, S. A. Mohamud, and E. N. Ceesay, “Transfer Learning Model for Disrupting Misinformation during a COVID-19 Pandemic,” 2021 IEEE 11th Annu. Comput. Commun. Work. Conf. CCWC 2021, pp. 245–250, 2021, doi: 10.1109/CCWC51732.2021.9376066.
J. H. Lee, “Understanding Public Attitudes Toward COVID-19 with Twitter,” 2021 IEEE Syst. Inf. Eng. Des. Symp. SIEDS 2021, pp. 1–6, 2021, doi: 10.1109/SIEDS52267.2021.9483708.
S. Vydra and J. Kantorowicz, “Tracing Policy-relevant Information in Social Media: The Case of Twitter before and during the COVID-19 Crisis,” Stat. Polit. Policy, vol. 12, no. 1, pp. 87–127, 2021, doi: 10.1515/spp-2020-0013.
M. S. Ahmed, T. T. Aurpa, and M. M. Anwar, “Detecting sentiment dynamics and clusters of Twitter users for trending topics in COVID-19 pandemic,” PLoS One, vol. 16, no. 8 August, pp. 1–20, 2021, doi: 10.1371/journal.pone.0253300.
S. Fazeli, D. Zamanzadeh, A. Ovalle, T. Nguyen, G. Gee, and M. Sarrafzadeh, “COVID-19 and Big Data: Multi-faceted Analysis for Spatio-temporal Understanding of the Pandemic with Social Media Conversations,” 2021.
N. Öztürk, H. Tozan, and Ö. Vayvay, “A new decision model approach for health technology assessment and a case study for dialysis alternatives in Turkey,” Int. J. Environ. Res. Public Health, vol. 17, no. 10, 2020, doi: 10.3390/ijerph17103608.
E. Delice, H. Tozan, M. A. Karadayi, M. Harnicarova, and B. Turan, “An Integrated Framework for Non-Traditional Machining Process Technology Selection in Healthcare Applications,” Teh. Vjesn., vol. 29, no. 6, pp. 2137–2146, 2022, doi: 10.17559/TV-20220503130337.
A. S. M. Masud and A. Ravi Ravindran, Multiple criteria decision making. 2008.
L. C. Jiang, T. H. Chu, and M. Sun, “Characterization of Vaccine Tweets During the Early Stage of the COVID-19 Outbreak in the United States: Topic Modeling Analysis,” JMIR Infodemiology, vol. 1, no. 1, p. e25636, 2021, doi: 10.2196/25636.
K. Thurson, “Parenting , Vaccines , and COVID-19 : A Machine-Learning Approach,” no. May, 2021.
N. S. Sattar and S. Arifuzzaman, “Covid-19 vaccination awareness and aftermath: Public sentiment analysis on twitter data and vaccinated population prediction in the usa,” Appl. Sci., vol. 11, no. 13, 2021, doi: 10.3390/app11136128.
S. Gbashi, O. A. Adebo, W. Doorsamy, and P. B. Njobeh, “Systematic delineation of media polarity on COVID-19 vaccines in Africa: Computational linguistic modeling study,” JMIR Med. Informatics, vol. 9, no. 3, 2021, doi: 10.2196/22916.
J. C. Lyu, E. Le Han, and G. K. Luli, “Covid-19 vaccine-related discussion on twitter: Topic modeling and sentiment analysis,” J. Med. Internet Res., vol. 23, no. 6, pp. 1–12, 2021, doi: 10.2196/24435.
S. W. H. Kwok, S. K. Vadde, and G. Wang, “Tweet topics and sentiments relating to COVID-19 vaccination among Australian twitter users: Machine learning analysis,” J. Med. Internet Res., vol. 23, no. 5, 2021, doi: 10.2196/26953.
A. Karami, M. Zhu, B. Goldschmidt, H. R. Boyajieff, and M. M. Najafabadi, “COVID-19 Vaccine and Social Media in the U.S.: Exploring Emotions and Discussions on Twitter,” Vaccines, vol. 9, no. 10, p. 1059, 2021, doi: 10.3390/vaccines9101059.
R. Proscia, “Sentiment Analysis on Twitter,” no. February, 2016.
R. Vording, “Harvesting unstructured data in heterogenous business environments ; exploring modern web scraping technologies,” pp. 1–9.
B. Naimi and A. Voinov, “StellaR: A software to translate Stella models into R open-source environment,” Environ. Model. Softw., vol. 38, pp. 117–118, 2012, doi: 10.1016/j.envsoft.2012.05.012.
S. Symeonidis, D. Effrosynidis, and A. Arampatzis, “A comparative evaluation of pre-processing techniques and their interactions for twitter sentiment analysis,” Expert Syst. Appl., vol. 110, pp. 298–310, 2018, doi: 10.1016/j.eswa.2018.06.022.
İ. Kızıltan, “An Evaluation of Istanbul Real Estate Posts Based on Text Analysis and Topic Modelling,” 2018.
S. Osailan, “Topic modeling location-based social media applications,” Diss. Abstr. Int. Sect. B Sci. Eng., vol. 81, no. 12-B, 2020.
M. Bhat, M. Kundroo, and T. A. Tarray, “Deep LDA : A new way to topic model Deep LDA : A new way to topic model,” no. July, 2019, doi: 10.1080/02522667.2019.1616911.
E. Ozkan, “Topic Modelling by Using Latent Dirichlet Allocation,” 2015.
I. Vayansky, S. A. P. Kumar, and Z. Li, “An Evaluation of Geotagged Twitter Data during Hurricane Irma Using Sentiment Analysis and Topic Modeling for Disaster Resilience,” Int. Symp. Technol. Soc. Proc., vol. 2019-Novem, pp. 0–5, 2019, doi: 10.1109/ISTAS48451.2019.8937859.
S. Kwon and A. Park, “Understanding user responses to the COVID-19 pandemic on Twitter from a terror management theory perspective: Cultural differences among the US, UK and India,” Comput. Human Behav., vol. 128, no. May 2021, p. 107087, 2022, doi: 10.1016/j.chb.2021.107087.
L. Zou and W. W. Song, “LDA-TM: A two-step approach to Twitter topic data clustering,” Proc. 2016 IEEE Int. Conf. Cloud Comput. Big Data Anal. ICCCBDA 2016, pp. 342–347, 2016, doi: 10.1109/ICCCBDA.2016.7529581.
D. M. Blei, A. Y. Ng, and M. I. Jordan, “Latent Dirichlet Allocation David,” J. Mach. Learn. Res., vol. 3, pp. 993–1022, 2003.
Z. A. Güven, R. T. Erdoğan, B. Diri, and T. Çakaloğlu, “n-seviyeli Gizli Dirichlet Ayırımı ile Türkçe Tivit Duygularının Sınıflandırılması Classification of TurkishTweet Emotions by n-stage Latent Dirichlet Allocation,” Academia.Edu, no. Asyu, 2012.
“Latent Dirichlet Allocation(LDA) Kullanarak Nasıl Topic Modeling Yapılır? | by Büşra Gökmen | Medium.” .
“Gizli Dirichlet Tahsisi: pyLDAvis ile sezgi, matematik, uygulama ve görselleştirme | tarafından Ioana | Veri Bilimine Doğru.” .
K. Para, “Investigation of the Causality Between Twitter Comments and Altcoin Cyptocurrency Market by Sentiment Analysis During the COVID-19 pandemic : Ripple Example,” no. September, pp. 362–381, 2021.
O. Kaynar, M. Yildiz, Y. Görmez, and A. Albayrak, “Makine Öğrenmesi Yöntemleri ile Duygu Analizi,” no. September, pp. 234–241, 2016.
A. Sarlan, C. Nadam, and S. Basri, “Twitter sentiment analysis,” Conf. Proc. - 6th Int. Conf. Inf. Technol. Multimed. UNITEN Cultiv. Creat. Enabling Technol. Through Internet Things, ICIMU 2014, pp. 212–216, 2015, doi: 10.1109/ICIMU.2014.7066632.

No competing interests reported.

Download PDF

Version 1

posted

You are reading this latest preprint version

Public Perception Towards Children's COVID-19 Vaccination with Natural Language Processing

Status:

Version 1

Abstract

Figures

1 Introduction

2 Background

2.1 COVID-19 echo and vaccine perception in the world

2.2 Natural language processing (NLP) studies in the COVID-19

2.2.1 Topic modeling and sentiment analysis studies on the perception of COVID-19

2.2.2 COVID-19 vaccine perception with topic modeling and sentiment analysis

3 Methodology

3.1 Web scraping and dataset generation

3.2 Pre-processing

3.3 Topic modelling

3.4 Sentiment Analysis

4 Results

5 Conclusion and Discussion

Declarations

References

Additional Declarations

Status:

Version 1