Tweet Text Research Articles

Analysis of tweets from 25 Unofficial ASCO Representatives from ASCO 2024

Aim of the studyThe American Society of Clinical Oncology (ASCO) hosts an annual conference and is one of the largest medical conferences globally. For ASCO 2024, they selected 25 featured voices, advertising them as individuals to follow on Twitter/X throughout the conference. The aim of this study was to characterize tweets from each of the 25 featured individuals. Brief summary of the methodsFrom May 16, 2024 through June 19, 2024 we filtered tweets by using the twitter advanced search tool and applying the following criteria in the tweet text: #ASCO OR #ASCO24 OR #ASCO2024. Tweets were classified as being about research (yes or no) and if media was included (yes or no). Additional information about research related tweets was gathered such as whether the research was observational or interventional. If it was interventional we characterized whether the tweet was supportive, critical, or neutral of the research. ResultsWe find that 229 (28.3%) of 810 tweets commented on research. Of the 229 research related tweets, 136 (59.4%) were related to interventional trials of which they were supportive 65.3% of the time. Media was included in 280 (34.6%) of 810 tweets. ConclusionIt is concerning that more tweets include media than are research related. And when research is referenced, individuals are rarely critical. Trials presented at ASCO have the ability to impact guidelines so it is important that they are appraised critically which can be difficult to complex methodology. This analysis provides data and brings awareness so that improvements can be made in future ASCO conferences. Policy summary statement at the endThe American Society of Clinical Oncology should disclose financial relationships in the future when selecting featured voices to represent the organization and provide insight.

Code-switching in South Asian English CMC

Abstract This paper analyses the use of indigenous language elements including code-switching in two contrasting genres, i.e. group chats and Twitter memes along with tweets, in the English communication of South Asian (Bangladeshi, Indian, Pakistani, and Sri Lankan) internet users. The results from group chats show that one of the most common lexical indigenous elements are tags, for example, address forms like yaar, machan, and da which all can be translated to English as ‘dude’ or ‘buddy’. The analysis of Twitter memes along with tweets shows that despite the tweet text being in English, the South Asian users tend to employ memes with indigenous text more often as compared to English memes for political satire. Overall, the study finds that code-switching and indigenous resources are used to create a sense of localness in English communication, whether it is group chats or Twitter memes and tweets.

Digital Epidemiology of Prescription Drug References on X (Formerly Twitter): Neural Network Topic Modeling and Sentiment Analysis.

Data from the social media platform X (formerly Twitter) can provide insights into the types of language that are used when discussing drug use. In past research using latent Dirichlet allocation (LDA), we found that tweets containing "street names" of prescription drugs were difficult to classify due to the similarity to other colloquialisms and lack of clarity over how the terms were used. Conversely, "brand name" references were more amenable to machine-driven categorization. This study sought to use next-generation techniques (beyond LDA) from natural language processing to reprocess X data and automatically cluster groups of tweets into topics to differentiate between street- and brand-name data sets. We also aimed to analyze the differences in emotional valence between the 2 data sets to study the relationship between engagement on social media and sentiment. We used the Twitter application programming interface to collect tweets that contained the street and brand name of a prescription drug within the tweet. Using BERTopic in combination with Uniform Manifold Approximation and Projection and k-means, we generated topics for the street-name corpus (n=170,618) and brand-name corpus (n=245,145). Valence Aware Dictionary and Sentiment Reasoner (VADER) scores were used to classify whether tweets within the topics had positive, negative, or neutral sentiments. Two different logistic regression classifiers were used to predict the sentiment label within each corpus. The first model used a tweet's engagement metrics and topic ID to predict the label, while the second model used those features in addition to the top 5000 tweets with the largest term-frequency-inverse document frequency score. Using BERTopic, we identified 40 topics for the street-name data set and 5 topics for the brand-name data set, which we generalized into 8 and 5 topics of discussion, respectively. Four of the general themes of discussion in the brand-name corpus referenced drug use, while 2 themes of discussion in the street-name corpus referenced drug use. From the VADER scores, we found that both corpora were inclined toward positive sentiment. Adding the vectorized tweet text increased the accuracy of our models by around 40% compared with the models that did not incorporate the tweet text in both corpora. BERTopic was able to classify tweets well. As with LDA, the discussion using brand names was more similar between tweets than the discussion using street names. VADER scores could only be logically applied to the brand-name corpus because of the high prevalence of non-drug-related topics in the street-name data. Brand-name tweets either discussed drugs positively or negatively, with few posts having a neutral emotionality. From our machine learning models, engagement alone was not enough to predict the sentiment label; the added context from the tweets was needed to understand the emotionality of a tweet.

Arabic Spam Tweets Classification: A Comprehensive Machine Learning Approach

Nowadays, one of the most common problems faced by Twitter (also known as X) users, including individuals as well as organizations, is dealing with spam tweets. The problem continues to proliferate due to the increasing popularity and number of users of social media platforms. Due to this overwhelming interest, spammers can post texts, images, and videos containing suspicious links that can be used to spread viruses, rumors, negative marketing, and sarcasm, and potentially hack the user’s information. Spam detection is among the hottest research areas in natural language processing (NLP) and cybersecurity. Several studies have been conducted in this regard, but they mainly focus on the English language. However, Arabic tweet spam detection still has a long way to go, especially emphasizing the diverse dialects other than modern standard Arabic (MSA), since, in the tweets, the standard dialect is seldom used. The situation demands an automated, robust, and efficient Arabic spam tweet detection approach. To address the issue, in this research, various machine learning and deep learning models have been investigated to detect spam tweets in Arabic, including Random Forest (RF), Support Vector Machine (SVM), Naive Bayes (NB) and Long-Short Term Memory (LSTM). In this regard, we have focused on the words as well as the meaning of the tweet text. Upon several experiments, the proposed models have produced promising results in contrast to the previous approaches for the same and diverse datasets. The results showed that the RF classifier achieved 96.78% and the LSTM classifier achieved 94.56%, followed by the SVM classifier that achieved 82% accuracy. Further, in terms of F1-score, there is an improvement of 21.38%, 19.16% and 5.2% using RF, LSTM and SVM classifiers compared to the schemes with same dataset.

Understanding the impact of geotagging on location inference models for accurate generalization to non-geotagged datasets

For a location inference model to be successful, the properties of the geotagged tweets where the location inference model is developed and those of the non-geotagged tweets where the location inference model is applied need to match. We investigated location mentions within the tweet text field of 3,953,166 geotagged and 2,783,609 non-geotagged tweets across five of the most prominent Twitter sources. Specifically, we compared the frequency and the location entity types used to infer the locations within the two datasets. Overall we found statistically significant differences in location mentions between the two datasets. However, although statistically significant, thirteen of the fifteen analysed location entities, showed low effect sizes. We conclude that location inference models trained on geotagged datasets can generalize a non-geotagged dataset if special adjustments are made on the development of the location inference models.

Acupuncture Topics on Twitter (Currently X) in English and Japanese: Co-occurrence Network Analysis.

Background With the growth of social media, there has been an increase in health-related studies utilizing data obtained from such sites and applications. Although acupuncture is used as a complementary alternative medicine worldwide, there is little research on acupuncture utilizing social media data. This study investigates the topics related to acupuncture on Twitter, currently known as X, in English and Japanese. Methods We collected tweets containing the English word "acupuncture" and its Japanese equivalent using Twitter's application programming interface from January 1, 2022to December 31, 2022. After extracting the top 50 frequently occurring words from the collected tweet texts separately for each language, we conducted a co-occurrence network analysis for those words, in order to evaluate the patterns of their occurrence. Results A total of 70,435 English tweets from 41,939 users and 188,671 Japanese tweets from 81,093 users were analyzed after excluding retweets and duplicate tweets. The co-occurrence network analysis revealed that topics related to pain, other complementary and alternative medicines, and acupuncture-related experiences were common in both languages. However, explanatory topics such as needle use, Chinese medicine, and body points were specific to English tweets, while those about beauty care were specific to Japanese tweets. Conclusion The number of tweets regarding acupuncture and the number of users who posted them were both higher in Japanese than in English. Some acupuncture topics were common in both languages, while other topics were specific to each language. The findings of this study provide valuable insights into understanding information about acupuncture on social media.

Deep Neural Networks for Location Reference Identification From Bilingual Disaster-Related Tweets

Twitter is increasingly being used during disasters to communicate with authorities, ascertain the ground reality, and coordinate real-time rescue and recovery activities. Geographical location information about users and events is critical in these scenarios. Geotagged tweets are extremely infrequent, and other location fields, such as user location and place name, are unreliable. The extraction of geographical information from tweet text is limited by the fact that individuals frequently publish multilingual tweets that contain numerous grammatical and spelling errors, as well as nonstandard acronyms. As a result, determining the geographical location of the tweet is a challenging problem. This article presents a technique based on deep neural networks for extracting geographical references mentioned in bilingual tweets. Several deep learning-based models, including convolutional neural networks (CNNs), long short-term memory (LSTM), bidirectional LSTM (Bi-LSTM), and attention-based Bi-LSTM, are implemented on real-world English and Hindi language tweets to determine their suitability for extracting location references. The proposed CNN, along with a conditional random field at the last layer, is found to perform better than other models, with an <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">$F_1$</tex-math> </inline-formula> -score of 0.858. The findings of this study can aid in early event detection, pinpointing the area of devastation and victims, real-time traffic management, and a number of other location-based applications. The suggested system’s code and trained model may be obtained at https://github.com/Abhinavkmr/Bi-lingual-location-reference-identification.git.

Urban Themes Constructed in the Persian Twitter

This study was conducted to investigate Persian tweets with urban content on Twitter. The main attention of this research is 22 city-related words applied to the tweets. The method of this research is to mine big data extracted from Twitter over the course of four weeks. The basis for analyzing is the text of tweets. The applied concepts and theories are borrowed from Habermas's public sphere theory, Lefebvre's spatial dialectic, and Manuel Castles’ space of flows. The results obtained from the analysis of tweets suggest that Persian Twitter users emphasize the central cities of Tehran and Isfahan when discussing the city. The space of flow theory is also visible without it contradicting the spatial dialectic of Lefebvre due to the spatial structure of Twitter, as users go back and forth between the urban space and the online space in an attempt to have a conversation with each other about urban issues. In Persian Twitter without historical and spatial logic, users write about the city and communicate with each other. Finally, despite the centrism mentality in Persian Twitter, it has challenged the possibility of forming a dialogue space for citizens.

Sarcasm‐based tweet‐level stress detection

AbstractPsychological stress has evolved as an important health concern across the globe. The vulnerability to stress and the ramifications of it have only worsened during the time of the COVID‐19 pandemic. This necessitates a timely diagnosis of stress before the condition progresses to chronicity. In this context, the popularity of social media like Twitter, where large numbers of users share opinions without any social stigma, has emerged as a major resource of human opinions. This has led to an increased research interest in social media‐based stress detection techniques. However, tweet‐level stress detection techniques in the literature have left a void in leveraging the text information in tweets, especially the presence of sarcastic expressions in the tweet's text content. To this end, a novel method called “Sarcasm‐based Tweet‐Level Stress Detection” (STSD) is proposed in this work with the modification of the logistic loss function to detect tweet‐level stress by availing the information of sarcasm that exists in the tweet‐content. The principle of the STSD model is to minimize the loss for non‐sarcastic tweets while maximising the loss for sarcastic tweets. Furthermore, extensive preprocessing and dimensionality reduction are performed using kernel principal component analysis (kernel PCA) to improve the performance by reducing the dimensions. The experimental results show that the proposed STSD model, when applied along with kernel PCA, records a significant improvement in accuracy by a minimum of 5.25% and a maximum of 9.19% over baseline models. Also, there is an increment in F1‐score by at least 0.085 points and a maximum of 0.164 points when compared to the baseline models.

Navigating the asthma network on Twitter: Insights from social network and sentiment analysis.

Asthma is a condition in which the airways become inflamed and constricted, causing breathing difficulties, wheezing, coughing, and chest tightness. Social networks can have a substantial effect on asthma management and results. However, no studies of social networks addressing asthma have been undertaken. The aim of this research was to identify the significant social network structures, key influencers, top topics, and sentiments of asthma-related Twitter conversations. All the tweets collected for this study included the keyword "asthma" or were mentioned in or in replies to tweets that were performed. For this study, a random sample of Twitter data was collected using NodeXL Pro software between December 1, 2022, and January 29, 2023. The data collected includes the user's display name, Twitter handle, tweet text, and the tweet's publishing date and time. After being imported into the Gephi application, the NodeXL data were then shown using the Fruchterman-Reingold layout method. In our study, SNA (Social Network Analysis) metrics were utilized to identify the most popular subject using hashtags, sentiment-related phrases (positive, negative, or neutral), and top influencer by centrality measures (degree, betweenness). The study collected 48,122 tweets containing the keyword "asthma" or mentioned in replies. News reporters and journalists emerged as top influencers based on centrality measures in Twitter conversations about asthma, followed by government and healthcare institutions. Education, trigger factors (e.g., cat exposure, diet), and associated conditions were highly discussed topics on asthma-related social media posts (e.g., sarscov2, copd). Our study's sentiment analysis revealed that there were 8427 phrases associated neutral comments (18%), 12,582 words reflecting positive viewpoints (26%), and 27,111 words reflecting negative opinions (56%). This study investigates the relevance of social media influencers, news reporters, health experts, health organizations, and the government in the dissemination and promotion of asthma-related education and awareness during public health information.

Emotion Detection using Twitter Datasets and Spacy Algorithm

People show emotions for everyday communication. Emotions are identified by facial expressions, behavior, writing, speaking, gestures and physical actions. Emotion plays a vital role in the interaction between two people. The detection of emotions through text is a challenge for researchers. Emotion detection from the text can be useful for real-world application. Automatic emotion detection in the original text aims to recognize emotions in any digital medium by using natural language processing techniques and different approaches. Enabling machines with the ability to recognize emotions in a particular kind of text such as twitter’s tweet has important applications in sentiment analysis and affective computing. We have worked on the newly published gold dataset (AIT-2018) and propose a model consisting of lexicalbased using WordNet-Affect and EmoSenticNet with supervised classifiers for detecting emotions in a tweet text.

Analyzing the Use of Social Media Data to Understand Long-Term Crisis Management Challenges of COVID-19

In the past three years, social media has had a significant impact on our lives, including crisis management. The COVID-19 pandemic highlighted the importance of accurate information and exposed the spread of false information. This paper specifically examines the COVID-19 crisis and analyzes relevant literature to provide insights for national authorities and organizations. Utilizing social media data for crisis management poses challenges due to its unstructured nature. To overcome this, the paper proposes a comprehensive method that addresses all aspects of long-term crisis management. This method relies on labeled and structured information for accurate sentiment analysis and classification. An automated approach is presented to annotate and classify tweet texts, reducing manual labeling and improving classifier accuracy. The framework involves generating topics using Latent Dirichlet Allocation (LDA) and ranking them with a new algorithm for data annotation. The labeled text is transformed into feature representation using Bert embeddings, which can be utilized in deep learning models for categorizing textual data. The primary aim of this paper is to offer valuable insights and resources to researchers studying crisis management through social media literature, with a specific focus on high-accuracy sentiment analysis.

Data and models for stance and premise detection in COVID-19 tweets: Insights from the Social Media Mining for Health (SMM4H) 2022 shared task

The COVID-19 pandemic has sparked numerous discussions on social media platforms, with users sharing their views on topics such as mask-wearing and vaccination. To facilitate the evaluation of neural models for stance detection and premise classification, we organized the Social Media Mining for Health (SMM4H) 2022 Shared Task 2. This competition utilized manually annotated posts on three COVID-19-related topics: school closures, stay-at-home orders, and wearing masks. In this paper, we extend the previous work and present newly collected data on vaccination from Twitter to assess the performance of models on a different topic. To enhance the accuracy and effectiveness of our evaluation, we employed various strategies to aggregate tweet texts with claims, including models with feature-level (early) fusion and dual-view architectures from the SMM4H 2022 Task 2 leaderboard. Our primary objective was to create a valuable dataset and perform an extensive experimental evaluation to support future research in argument mining in the health domain.

Penerapan Algortitma C4.5 untuk Klasifikasi Sentimen Masyarakat terhadap #RUUKUHP pada Twitter

Social media, especially Twitter, has developed into an important tool for people to share their opinions and feelings widely. Users often use hashtags to share messages related to certain topics. Some of the issues that lead to the need for sentiment analysis of the Draft Criminal Code are social impact, Public disapproval, Potential legal uncertainty, Potential abuse, Support and criticism. By conducting a sentiment analysis of the draft Penal Code, the government and policymakers can better understand the views of the public, identify possible problems and address them, and make necessary improvements or clarifications to the draft law. This can help ensure that the draft Penal Code has greater public support and adheres to good legal principles. The classification of public responses to this hashtag provides a significant snapshot of public attitudes and perspectives. This study aims to classify public sentiment towards the RUUKUHP hashtag on the Twitter platform using the C4.5 algorithm. This study uses a collection of tweets with the hashtag RUUKUHP which are manually categorized into two and three sentiment categories, namely positive, negative and positive, negative and neutral. In this study, data preprocessing is carried out before training the model which includes removing links, special characters, removing stopwords, and word tokenization. Furthermore, this research uses text representation methods such as TF-IDF to extract features from the tweet text and convert them into numerical vectors used by the C4.5 algorithm. After training the classification model using the C4.5 algorithm with the classified dataset, it evaluates the performance of the model with the metrics of accuracy, recall, precision, and F1 score. Experimental results using 2 categories of Negative and Positive show that the model applied with the C4.5 algorithm achieved an accuracy of 96.6% with a recall of 96.6%, a percision of 97.1% and an F1 score of 96.8. And experiments using 3 categories of Negative, Positive and Neutral achieved an accuracy of 67%, a recall of 67%, a precision of 65%, and an F1 score of 66%. Thus it can be concluded that the results of the RUUKUHP hashtag sentiment classification with 2 class predictions are more relevant than 3 sentiment class predictions with a value reaching 96.6%.

Emoji, Sentiment and Emotion Aided Cyberbullying Detection in Hinglish

The advent of the Internet is a boon to society. However, many of its banes cannot be undermined, cyberbullying being one of them. The emotional state and sentiment of a person have a significant influence on the intended content. The current work is the first attempt in investigating the role of sentiment and emotion information for identifying cyberbullying in the Indian scenario. From Twitter, a benchmark Hind&#x2013;English code-mixed corpus called BullySentEmo has been developed as there is no dataset available labeled with bully, sentiment, and emotion. Moreover, emoji information available with tweet texts can provide better understanding of user intention. The developed dataset consists of both modalities, tweet text, and emoji. In India, the majority of communication on different social media platforms is based on Hindi and English and language switching is a common practice in digital communication. An attention-based multimodal, adversarial multitasking framework is proposed for cyberbully detection (CBD) considering two auxiliary tasks: sentiment analysis (SA) and emotion recognition (ER). Experimental results indicate that compared to unimodal and single-task variants, the proposed framework improves the performance of the main task, i.e., CBD, by 3.59% and 2.56% in terms of accuracy and F1-Score, respectively. Furthermore, two different benchmark datasets (Twitter dataset and Aggression dataset) have been considered to show the robustness of our proposed model.

Using Machine Learning to Classify Information Related to Child Rearing of Infants from Twitter

It is difficult to obtain necessary information accurately from Social Networking Service (SNS) while raising children, and it is thought that there is a certain demand for the development of a system that presents appropriate information to users according to the child's developmental stage. There are still few examples of research on knowledge extraction that focuses on childcare. This research aims to develop a system that extracts and presents useful knowledge for people who are actually raising children, using texts about childcare posted on Twitter. In many systems, numbers in text data are just strings like words and are normalized to zero or simply ignored. In this paper, we created a set of tweet texts and a set of profiles created according to the developmental stages of infants from "0-year-old child" to "6-year-old child". For each set, we used ML algorithms such as NB (Naive Bayes), LR (Logistic Regression), ANN (Approximate Nearest Neighbor algorithms search), XGboost, RF (random forest), decision trees, and SVM (Support Vector Machine) to compare with BERT (Bidirectional Encoder Representations from Transformers), a neural language model, to construct a classification model that predicts numbers from "0" to "6" from sentences. The accuracy rate predicted by the BERT classifier was slightly higher than that of the NB, LR, and ANN, XGboost, and RF, decision trees and SVM classifiers, indicating that the BERT classification method was better.

Clustering Content Types and User Roles Based on Tweet Text Using K-Medoids Partitioning Based

In this modern era, the spread of information occurs rapidly through social media. One of the channels for disseminating information is through the Twitter platform. Many Twitter users respond to existing content with positive, negative and neutral responses. One of the hot content to respond to is political content. This content is currently being discussed considering the approaching election of the 2024 Presidential Candidate of the Republic of Indonesia. One of the candidate pairs discussed was Anies Baswedan. With so many responses from Twitter users, it will be difficult to track whether users support Anies Baswedan to run as a presidential candidate due to the large number of responses. This study aims to determine the response of twitter users to the advancement of Anies Baswedan as a presidential candidate. The method used in this study is the K-Medoids Partitioning-Based algorithm based on twitter user text. This algorithm was chosen because it is easy to implement considering the basis of K-Medoids development is the K-Means algorithm but the K-Medoids algorithm can overcome the shortcomings of the K-Means algorithm which is sensitive to outliners. The evaluation will be done using Silhouette Score which produces a value of 0.35 with the number of clusters is 2. Then an analysis of each cluster is carried out by looking at the words in the cluster. As a result, from the two clusters formed, both clusters contain positive content and show that Twitter users support Anies Baswedan to run as a 2024 presidential candidate.

Knowledge-enhanced Prompt-tuning for Stance Detection

Investigating public attitudes on social media is important in opinion mining systems. Stance detection aims to analyze the attitude of an opinionated text (e.g., favor, neutral, or against) toward a given target. Existing methods mainly address this problem from the perspective of fine-tuning. Recently, prompt-tuning has achieved success in natural language processing tasks. However, conducting prompt-tuning methods for stance detection in real-world remains a challenge for several reasons: (1) The text form of stance detection is usually short and informal, which makes it difficult to design label words for the verbalizer. (2) The tweet text may not explicitly give the attitude. Instead, users may use various hashtags or background knowledge to express stance-aware perspectives. In this article, we first propose a prompt-tuning-based framework that performs stance detection in a cloze question manner. Specifically, a knowledge-enhanced prompt-tuning framework (KEprompt) method is designed, which consists of an automatic verbalizer (AutoV) and background knowledge injection (BKI). Specifically, in AutoV, we introduce a semantic graph to build a better mapping from the predicted word of the pretrained language model and detection labels. In BKI, we first propose a topic model for learning hashtag representation and introduce ConceptGraph as the supplement of the target. At last, we present a challenging dataset for stance detection, where all stance categories are expressed in an implicit manner. Extensive experiments on a large real-world dataset demonstrate the superiority of KEprompt over state-of-the-art methods.

Incomplete Gamma Integrals for Deep Cascade Prediction Using Content, Network, and Exogenous Signals

The behavior of information cascades (such as retweets) has been modeled extensively. While point process-based generative models have long been in use for estimating cascade growths, deep learning has greatly enhanced the integration of diverse features and signals. We observe two significant temporal signals in cascade data that have not been reported or exploited to our knowledge. First, the popularity of the cascade root is known to influence cascade size strongly; but we find that the effect falls off rapidly with time. Second, we find a measurable positive correlation between the novelty of the root content (with respect to a streaming external corpus) and the relative size of the resulting cascade. Responding to these observations, we propose <monospace>GammaCas</monospace>, a new cascade growth model as a parametric function of time, which combines deep influence signals from content (e.g., tweet text), network features (e.g., followers of the root user), and exogenous event sources (e.g., online news). Specifically, our model processes these signals through a customized recurrent network, whose states then provide the parameters of the cascade rate function, which is integrated over time to predict the cascade size. The network parameters are trained end-to-end using observed cascades. <monospace>GammaCas</monospace> outperforms seven recent and diverse baselines significantly on a large-scale dataset of retweet cascades coupled with time-aligned online news &#x2014; it beats the best baseline with 18.98&#x0025; increase in terms of Kendall&#x0027;s <inline-formula><tex-math notation="LaTeX">$\tau$</tex-math></inline-formula> correlation and a reduction of 19.2 in Mean Absolute Percentage Error. Extensive ablation and case studies unearth interesting insights regarding retweet cascade dynamics.

A hybrid approach based on linguistic analysis and fuzzy logic to ensure the surveillance of people having paranoid personality disorder towards Covid-19 on social media

This paper presents a supervised learning method for paranoid detection in French tweets. A classifier uses four groups of features (textual, linguistic, meta-data, timeline) that exploit a hybrid approach. This approach uses information obtained from the text of tweets by applying Natural Language Processing (NLP) techniques to analyse them, such as morphological analysis, syntactic analysis and sentence embedding. Thus, information about the user such as the number of followers and the number of shared posts. Besides, information about tweets such as the number of symbols and the number of hashtags. Moreover, information about the publication date of tweets such as the number of postings in the morning. Finally, statistical techniques to combine and filter the different types of features extracted from the previous steps in order to calculate the distance between the training corpus (the labelled data) and the test corpus (unlabelled data). In addition, the state mentioned statistical techniques are used for detecting the writing style of patients. In general, our method benefits from different types of features and preserves the principle of relativity by taking advantage of fuzzy logic. Our results are encouraging with an accuracy of 78% for the detection of paranoid people and 70% for the detection of the behaviour of these people towards Covid-19.