Improved graph convolutional network for emotion analysis in social media text

The Impact of Arabic Part of Speech Tagging on Sentiment Analysis: A New Corpus and Deep Learning Approach

This thesis examines the issue of ill-formed text (microtext) and shows its impact on sentiment analysis. The source of microtext is Twitter, WhatsApp, Facebook and other social media platforms. Words or phrases not in their standard language format like “lol” (laugh out loud), “c u 2nite” (see you tonight) and “plz” (please) are called out of vocabulary (OOV) terms. On the other hand, words or phrases in their standard forms, for example, “tonight”, “love”, or “talk to you later”, are called In Vocabulary (IV) terms. The motivation behind microtext normalization is its growing presence on both online and offline platforms. Microtext might appear to be an ill-formed text, which requires find-and-replace methods. Traditionally, in a Natural Language Processing (NLP) context, microtexts such as abbreviations and length shortening or extending are normalized, but phonemic and graphemic variation is often ignored. However, the features of microtext can be traced back to social, cognitive and geographic influences. To understand social media text, it is essential to develop a microtext normalization module for traditional and modern NLP models. The first section of this thesis focuses on combining unsupervised learning methods in 2 different chapters, 4 and 5. The first method includes lexicon creation to transform the most frequent microtexts and emoticons on social media and applies them to sentiment analysis. This method shows an accuracy increase of 4% when microtext normalization is applied before sentiment analysis. This method is extended with an application to a pre-trained chatbot to improve the chatbot’s understanding of microtext (social media slangs). This method shows a mean BLEU score of 0.8 for the SMS and Tweet dataset. The second method incorporates a phonetic based approach to transform any microtext into its phonetic equivalent before 1normalizing it. This method shows an accuracy improvement of 6% for sentiment analysis on SenticNet. The second part of the unsupervised learning is in Chapter 5, introducing the IPA-based method for microtext normalization and sentiment analysis. Epitran was used to transform a word to its International Phonetic Alphabet (IPA) equivalent. The International Phonetic Alphabet (IPA) is an alphabetic system of phonetic notation based primarily on the Latin script. There is a significant improvement in microtext normalization task and sentiment analysis over Soundex. The results also show that there is very little redundancy when words are transformed to phonetics in IPA. In addition, the accuracy of polarity detection using this method has an improvement over baseline by 5%. Chapter 6 discusses a simple sequence-to-sequence based Deep Learning method. As the language is dynamic, it is not easy to maintain the lexicon. Deep Learning alleviates the static lexicon with a probabilistic model to transform microtext into its standard form. The findings suggest that a semiotic class of microtext was easy to transform, whereas the phonetic class of microtext was not handled easily. The proposed model improves the pre-trained sentiment analysis model by 4%. We also introduce a corpus for English microtext normalization. The corpus contains microtext-containing sentences along with their correctly spelt sentences and their correct polarity. This dataset can tackle two NLP problems, i.e., microtext normalization and sentiment analysis. This corpus consists of 38% positive, 39.9% negative and 22.1% neutral. A pre-trained model shows an accuracy of 46.4 % on OOV text for sentiment analysis, showing considerable room for improvement. In summary, this thesis introduces several approaches to microtext normalization analysis and focuses on sentiment analysis application. Microtext cannot be thought of as a one-dimensional problem to NLP. It encapsulates many factors like regional (geographical), ethnic (national and racial), and social (class, age, gender, socioeconomic status and education), which cannot be ignored given the growing importance of social media in our daily life.

The exponential growth of user-generated content on social media platforms (e.g., Twitter, Weibo, Facebook) has created massive datasets rich in public sentiment. While sentiment analysis proves vital for understanding social trends, brand perception, and political developments, traditional methods struggle to handle the informal nature, noise, and context-dependent characteristics of social media language. This necessitates advanced computational technologies to extract meaningful insights. This paper summarizes the application of Python in social media text sentiment analysis, covering various methods supported by its NLP library (NLTK, TextBlob), machine learning/deep learning framework (TensorFlow, PyTorch), and discusses cross-platform comparative analysis and specific domain adaptation. The results indicate that Python-based models achieve high levels of accuracy. For example, the RoBERTa-BiLSTM-MHA model achieved an accuracy of 93.44%, and these models outperform conventional tools by 612% in terms of F1 scores. Key findings reveal cultural and platform-based disparities in sentiment expression. Specifically, Chinese social media platforms like Weibo emphasize economic sentiment, while Western platforms such as Twitter focus more on technical and ethical implications. The integration of sentiment dictionaries and multimodal data, including emojis, further enhances the robustness of these sentiment analysis models. Overall, this review underscores Pythons versatility in enabling scalable and real-time sentiment analysis, which in turn drives innovations in NLP research and practical applications.

Sentiment analysis is the process of computationally recognizing and classifying the attitudes conveyed in each text towards a particular topic, product, etc. which is either positive or negative. Sentiment analysis is one of the interesting applications of natural language processing (NLP) and which is used to analyze the social media. Text in social media is casual and it can be written either in code-switch or monolingual text. Several researchers have implemented sentiment analysis on monolingual text, though sentiments can be expressed in code-switch text. Sentiment analysis can be applied through deep learning (DL), machine learning (ML), or a Lexicon-based approach. Machine learning (ML) and deep learning (DL) methods are time-consuming, computationally expensive, and need training data for analysis. Lexicon-based method does not require training data and requires less time to find the sentiments in comparison with ML and DL. In this paper, we propose the Lexicon-based approach (NBLex) to analyze the sentiments expressed in Kannada-English code-switch text. This is the first effort that targets to perform sentiment analysis in Kannada-English code-switch text using the Lexicon-based approach. The proposed approach performed with better accuracy of 83.2% and 83% of F1-score.

BackgroundAlthough at present there is broad agreement among researchers, health professionals, and policy makers on the need to control and combat health misinformation, the magnitude of this problem is still unknown. Consequently, it is fundamental to discover both the most prevalent health topics and the social media platforms from which these topics are initially framed and subsequently disseminated.ObjectiveThis systematic review aimed to identify the main health misinformation topics and their prevalence on different social media platforms, focusing on methodological quality and the diverse solutions that are being implemented to address this public health concern.MethodsWe searched PubMed, MEDLINE, Scopus, and Web of Science for articles published in English before March 2019, with a focus on the study of health misinformation in social media. We defined health misinformation as a health-related claim that is based on anecdotal evidence, false, or misleading owing to the lack of existing scientific knowledge. We included (1) articles that focused on health misinformation in social media, including those in which the authors discussed the consequences or purposes of health misinformation and (2) studies that described empirical findings regarding the measurement of health misinformation on these platforms.ResultsA total of 69 studies were identified as eligible, and they covered a wide range of health topics and social media platforms. The topics were articulated around the following six principal categories: vaccines (32%), drugs or smoking (22%), noncommunicable diseases (19%), pandemics (10%), eating disorders (9%), and medical treatments (7%). Studies were mainly based on the following five methodological approaches: social network analysis (28%), evaluating content (26%), evaluating quality (24%), content/text analysis (16%), and sentiment analysis (6%). Health misinformation was most prevalent in studies related to smoking products and drugs such as opioids and marijuana. Posts with misinformation reached 87% in some studies. Health misinformation about vaccines was also very common (43%), with the human papilloma virus vaccine being the most affected. Health misinformation related to diets or pro–eating disorder arguments were moderate in comparison to the aforementioned topics (36%). Studies focused on diseases (ie, noncommunicable diseases and pandemics) also reported moderate misinformation rates (40%), especially in the case of cancer. Finally, the lowest levels of health misinformation were related to medical treatments (30%).ConclusionsThe prevalence of health misinformation was the highest on Twitter and on issues related to smoking products and drugs. However, misinformation on major public health issues, such as vaccines and diseases, was also high. Our study offers a comprehensive characterization of the dominant health misinformation topics and a comprehensive description of their prevalence on different social media platforms, which can guide future studies and help in the development of evidence-based digital policy action plans.

INTRODUCTION. Building effective communication with the target audience through media texts is a relevant technique of interactive interaction. The correlation of methods of interactive communication with a regional audience and the functional purpose of elements of a social media media text is studied. The purpose of the study is to substantiate the convergence of the functionality of interactive methods of communication with a regional audience in media texts of social media. The theoretical novelty of the work lies in identifying the convergence of the functionality of media texts, in introducing the concept of “landing” of media content into scientific circulation.MATERIALS AND METHODS. The theoretical basis of the study was the works of L.R. Duskaeva, E.A. Zvereva, A.A. Kazakov, N.I. Klushina, A.M. Shesterina on the use of interactive techniques in media texts. The most comprehensive Tambov communities on VKontakte and Telegram channels were selected as an empirical base. The study used the method of content analysis of social media texts for 2023-2024.RESULTS AND DISCUSSION. The following conclusions were formulated in the course of the study: 1) each speech technique performs several functions, which indicates the convergence of the functionality of media texts; 2) speech techniques of interactive communication in synthesis with technical means demonstrate effectiveness; 3) speech techniques are a manipulative means, and technical means are used to implement media communication tasks.CONCLUSION. Interactive speech techniques in media texts allow to influence the user, involve him in the news item and form a certain opinion about it, motivate to leave a social reaction to the media content. This gives the authors the opportunity to make the platform more recognizable and interesting, as well as to increase the effectiveness of the community.

Interaction via social media involves frequent code‐mixed text, spelling errors and noisy elements, which creates a bottleneck in the performance of natural language processing applications. This proposed work is the first approach for code‐mixed Hindi‐English social media text that comprises language identification, detection and correction of non‐word (Out of Vocabulary) errors as well as real‐word errors occurring simultaneously. Each identified language (Devanagari Hindi, Roman Hindi, and English) has its own complexities and challenges. Errors are detected individually for each language and a suggestive list of the erroneous words is created. After this, a fuzzy graph between different words of the suggestive lists is generated using various semantic relations in Hindi WordNet. Word embeddings and Fuzzy graph‐based centrality measures are used to find the correct word. Several experiments are performed on different social media datasets taken from Instagram, Twitter, YouTube comments, Blogs, and WhatsApp. The experimental results demonstrate that the proposed system corrects out‐of‐vocabulary words as well as real‐word errors with a maximum recall of 0.90 and 0.67, respectively for Dev_Hindi and 0.87 and 0.66, respectively for Rom_Hindi. The proposed method is also applied for state‐of‐art sentiment analysis approaches where the F1‐score has been visibly improved.

Background Health misinformation disseminated on social media detrimentally affects the public's reactions toward public health measures. This has led to costly and harmful public health crises. Due to a lack of a comprehensive expert system that collects and analyzes large volumes of social media data, public health officials are incapable of mitigating health misinformation trends online. This study aimed to remedy this by designing and developing a big data pipeline and ecosystem (Health Misinformation Analysis System (HMAS)) capable of identifying and analyzing health misinformation online. Methods The HMAS system relies on Python, Elastic Stack, and the Twitter V2 API as the main technologies forming the backbone of the system. It comprises of five main components: (1) Data Extraction Framework (DEF); (2) Latent Dirichlet Allocation Topic Model; (3) Sentiment Analyzer; (4) Misinformation Classifier; (5) Elastic Cloud Deployment. Components 1-4 are hosted on a virtual machine with low computing and memory requirements. Through the DEF, HMAS extracts data from the Twitter V2 API. The HMAS expert system uses pre-trained models to perform automatic health misinformation analysis. The analyzed data are loaded and visualized in Elastic Cloud through dashboards and analytics. Results The HMAS system performance is accurate and efficient. It has been successfully utilized by independent investigators to extract significant insights, including a fluoride-related health misinformation use case, which analyzed data from 2015 to 2021. Moreover, HMAS has been applied to a vaccine hesitancy use case (2007-2022) and a heat-related illnesses use case (2011-2022). Conclusions The novel HMAS expert system can potentially aid public health officials globally to detect and analyze concerning trends in health misinformation online. This system can also grow to integrate social media data from several platforms for multiplatform analysis and support non-Western language content. Key messages • The novel HMAS pipeline can collect, detect, and analyze the rapidly increasing amounts of misinformation proliferating on social media related to a particular topic or set of related topics. • Identifying and monitoring health misinformation trends prevalent on social media through analytics dashboards and anomaly/risk detection alerts can lead to prompt governmental intervention.

Developing Resources For Sentiment Analysis Of Informal Arabic Text In Social Media

The broad use of social media nowadays has led many users to express their opinions on various subjects through them. The need for these opinions to be automatically labeled and categorized according to their sentiment, has also arisen. In this paper, a novel sentiment analysis approach is introduced, which takes into account the total number of idiomatic expressions and emoticons that are used in the text, and simultaneously processes the original text in Greek along with its automatic translation in English as well. Moreover, the novelty of the proposed solution lies in the difficulty of Modern Greek language and the fact that the text in social media is mainly unstructured. The proposed methodology is tested on two distinct data sets of opinions regarding a certain matter, which have been collected from Facebook and Twitter respectively. Finally, we discuss the performance of various classiffication algorithms and we compare the extracted experimental results.

With the rapid expansion of user-generated content on social media platforms like Twitter, Facebook, and Reddit, accurately identifying sentiment from textual data has become an essential yet challenging task due to the informal, noisy, and contextually diverse nature of these platforms. To address this, we propose a Hybrid Attention-Driven Recurrent Neural Network (HA-RNN) model that effectively combines Bidirectional Gated Recurrent Units (Bi-GRU) with a sophisticated attention mechanism for sentiment classification. The model utilizes pre-trained GloVe embeddings (300 dimensions) to capture rich semantic features from raw text, enhancing the initial representation of social media data. The Bi-GRU layers are employed to model sequential dependencies Safeyah Tawil Department of Computer Science and Engineering, Faculty of Information Technology, Zarqa University, Zarqa, Jordan. University of Business and Technology, Jeddah, Saudi Arabia stawil@zu.edu.jo I. INTRODUCTION Social media platforms have become primary channels for individuals to express opinions, emotions, and sentiments on a wide range of topics, including politics, products, services, and global events. The explosive growth of platforms such as Twitter, Facebook, and Instagram has led to an overwhelming amount of unstructured textual data that offers valuable insights into public sentiment [1]. Analyzing this vast content can support businesses, governments, and researchers in understanding user perceptions, improving services, and detecting social trends. in both forward and backward directions, ensuring a comprehensive understanding of context within a sentence. The integrated attention layer enables the model to dynamically focus on sentiment-bearing words, thereby improving classification accuracy and interpretability. We evaluated the proposed model on two widely recognized datasets: the Twitter US Airline Sentiment Dataset and the Sentiment140 Dataset. The HA-RNN achieved an accuracy of 90.8% on the Twitter US Airline dataset and 88.5% on Sentiment140, outperforming traditional models such as CNN (84.3% accuracy), LSTM (86.7%), and Bi-GRU without attention (87.1%). Furthermore, the attention mechanism provided insightful visualization, highlighting the critical words influencing sentiment predictions. The model demonstrated a balanced performance with high precision, recall, and F1-scores, validating its robustness across different sentiment classes. Overall, the HA- RNN model presents an effective and interpretable solution for sentiment analysis on noisy and diverse social media texts, supporting applications in social monitoring, brand analysis, and opinion mining.

Sentiment analysis is the computational study of attitudes, opinions, and sentiments towards certain issues, products, individuals, and organizations. Companies and customers are making decisions by seeking opinions from social media text. Sentiment analysis is getting intelligent with the advancement of artificial intelligence and natural language processing. With a stunning increase in the use of social media, a huge volume of text available on these platforms is in imperfect and informal languages like Roman Urdu mixed with the English language. Present sentiment analysis techniques do not perform precisely on these code-mixed imperfect, informal, and poorly resourced languages. A promising solution is the use of deep learning models on these code-mixed Roman Urdu and English text. Therefore, the objective of this paper is to perform a sentiment analysis of code-mixed Roman Urdu and English social media text using state-of-the-art deep learning models. Our work is independent of lexical normalization, language dictionary, and code transfer indication. We perform sentiment analysis using Multilingual BERT (mBERT) and XLM-RoBERTa (XLM-R) models. The results reveal that performance of XLM-R model with tuned hyperparameters for code-mixed Roman Urdu and English social media text is better than the mBERT model with F1 score of 71%.

Social media platforms have become central spaces for young people to engage with a vast array of digital content, significantly influencing their identities, beliefs, and behaviors. By examining think‐aloud and interview data, this study investigated how adolescents conceptualize social media as sources of news, political information, and tools for identity negotiation. The results informed the creation of a conceptual map outlining how the process of information‐seeking occurs through social media texts. Findings revealed a nuanced relationship between young people and social media, with participants conceptualizing social media texts as poor quality, but engaging frequently. While young people demonstrated some awareness of the need to verify content and awareness of algorithmic influences shaping their feeds, they often failed to critically engage with the content, which limited their ability to form a comprehensive understanding of complex issues, potentially hindering their capacity for informed discourse. Data revealed that young people passively interacted with social media texts through the algorithm, though they engaged in reciprocal algorithmic manipulation, where they consciously curated their algorithmic output. This research offers critical insights for educators, policymakers, and social media platforms aiming to foster media literacy and promote informed, responsible digital citizenship amongst youth. Ultimately, the findings emphasize the need for media literacy education to support critical thinking and lateral reading skills, empowering youth to navigate the digital information landscape effectively.

Social networking sites have become an integral part of everyday life, where people interact, cooperate and quarrel with each other. Social media also encourages them to express their opinions and share their comments about their lives’ events or about the product they use. Opinions can be direct without any comparison (I like ABC phone) or they can be comparative (X-phone’s camera is better than Y-phone). Comparative opinions are useful in many applications, e.g. marketing intelligence, product benchmarking, and e-commerce. The automatic mining of comparative opinions is an important text mining problem and an area of increasing interest for different languages. This paper focuses on identification of comparative sentence from non-comparative ones in Arabic texts. A corpus was developed consisting of YouTube comments. This paper describes research experiments that aimed to apply data/text mining algorithms, natural language processing and linguistic classification for Arabic comparative text discovery. The results of these experiments along with the analysis are also presented. The results were promising reaching to 91% accuracy for the detection of comparative opinions.

Text emanated from users’ posts and comments on social media constitutes important piece of information for wide ranging Natural Language Processing (NLP) applications, such as Sentiment Analysis, Sarcasm Detection, Named Entity Identification, Question Answering and Information Retrieval (IR). Part–of–Speech (POS) tagging, a prerequisite for all such applications, augments tag information to the raw text. However, an inherent tendency of social media users to include multilingual contents in their posts, called code-mixing, poses challenge to POS tagging. Besides, intricate and free style writing add to the complexity of problem. To cope with the issue, a Hidden Markov Model (HMM) based supervised algorithm has been introduced for POS tagging of code–mixed Indian social media text. Publicly available social media text of Indian Languages (ILs), particularly English, Hindi, Bengali and Telugu, have been used to train and test the proposed system. Correctness of system annotated tags has been evaluated on ground of F-measure.