Supervised Sentiment Analysis Research Articles

Supervised Sentiment Analysis of COVID-19 Tweets: Insights into Public Reactions and Trends

In the modern digital era, social media platforms have experienced exponential growth, serving as a medium for individuals to share their emotions and opinions. Twitter, a prominent social media platform, hosts a substantial number of users who regularly express their thoughts through tweets. This research paper focuses on conducting sentiment analysis on tweets pertaining to the COVID-19 situation in India. By utilizing a lexicon-based approach using TextBlob, we determine sentiment scores for each tweet. These scores are then used to categorize the tweets into Positive, Negative, or Neutral sentiment categories. To further enhance the accuracy of sentiment classification, we employ the Naïve Bayes machine learning model, achieving an impressive accuracy rate of 87.65%. The sentiment analysis results obtained through this study provide valuable insights for governmental bodies and other organizations, enabling them to formulate effective policies in order to address similar issues in the future.

Practical Sentiment Analysis for Education: The Power of Student Crowdsourcing

Sentiment analysis provides a promising tool to automatically assess the emotions voiced in written student feedback such as periodically collected unit-of-study reflections. The commonly used dictionary-based approaches are limited to major languages and fail to capture contextual differences. Pretrained large language models have been shown to be biased and online versions raise privacy concerns. Hence, we resort to traditional supervised machine learning (ML) approaches which are designed to overcome these issues by learning from domain-specific labeled data. However, these labels are hard to come by -- in our case manually annotating student feedback is prone to bias and time-consuming, especially in high-enrollment courses. In this work, we investigate the use of student crowdsourced labels for supervised sentiment analysis for education. Specifically, we compare crowdsourced and student self-reported labels with human expert annotations and use them in various ML approaches to evaluate the performance on predicting emotions of written student feedback collected from large computer science classes. We find that the random forest model trained with student-crowdsourced labels tremendously improves the identification of reflections with negative sentiment. In addition to our quantitative study, we describe our crowdsourcing experiment which was intentionally designed to be an educational activity in an introduction to data science course.

Semi-supervised learning for sentiment classification with ensemble multi-classifier approach

Supervised sentiment analysis ideally uses a fully labeled data set for modeling. However, this ideal condition requires a struggle in the label annotation process. Semi-supervised learning (SSL) has emerged as a promising method to avoid time-consuming and expensive data labeling without reducing model performance. However, the research on SSL is still limited and its performance needs to be improved. Thus, this study aims to create a new SSL-Model for sentiment analysis. The Ensemble Classifier SSL model for sentiment classification is introduced. The research went through pre-processing, vectorization, and feature extraction using TF-IDF and n-grams. Support Vector Machine (SVM) or Random Forest for tokenization was used to separate unigram, bigram, and trigram in model generation. Then, the outputs of these models were combined using stacking ensemble approach. Accuracy and F1-score were used for the evaluation. IMDB datasets and US Airlines were used to test the new SSL models. The conclusion is that the sentiment annotation accuracy is highly dependent on the suitability of the dataset with the machine learning algorithm. In IMDB dataset, which consists of two classes, it is better to use SVM. In the US Airlines consisting of three classes, SVM is better at improving the model performance against the baseline, but RF is better at achieving the baseline performance even though it fails to maintain the model performance.

Dual sentiment analysis with two sides of one review for polarity shift

The Bag Of Words' performance is limited because to inherent flaws in its treatment of the polarity shift problem; the bag of words representation considers the two opposite texts to be quite similar. To address this issue, a model known as dual sentiment analysis (DSA) has been developed. Dual Sentiment Analysis expands sentiment categorization analysis from one side of a review to two sides of a single review. For both training and test reviews, a data expansion strategy is presented to solve sentiment reversed review. In the sentiment classifier, a dual training method is utilized to train the original and reversed reviews. The Dual Sentiment Analysis paradigm divides polarity categorization into three categories (positive, negative, and neutral) (positive, negative and neutral). A corpus-based method for constructing a pseudo-antonym dictionary is devised to remove dual sentiment analysis' reliance on an external antonym dictionary for review reversal. The findings show that DSA is effective in supervised sentiment analysis.

A Japanese Subjective Well-Being Indicator Based on Twitter Data

Abstract This study presents for the first time the SWB-J index, a subjective well-being indicator for Japan based on Twitter data. The index is composed by eight dimensions of subjective well-being and is estimated relying on Twitter data by using human supervised sentiment analysis. The index is then compared with the analogous SWB-I index for Italy in order to verify possible analogies and cultural differences. Further, through structural equation models, we investigate the relationship between economic and health conditions of the country and the well-being latent variable and illustrate how this latent dimension affects the SWB-J and SWB-I indicators. It turns out that, as expected, economic and health welfare is only one aspect of the multidimensional well-being that is captured by the Twitter-based indicator.

HyVADRF: Hybrid VADER–Random Forest and GWO for Bitcoin Tweet Sentiment Analysis

In recent years, Bitcoin and other cryptocurrencies have been increasingly considered an investment option for emerging markets. However, its erratic behavior has discouraged some potential investors. To get insights into its behavior and price fluctuation, past studies have discovered the correlation between Twitter sentiments and Bitcoin behavior. Most of them have focused exclusively on their relationships, instead of the Twitter sentiment analysis itself. Finding the most suitable classification algorithms for sentiment analysis for this kind of data is challenging. For enormous data of Twitter, unlabeled data can be time-consuming and expensive for the supervised sentiment analysis approach, which has been studied to be superior to unsupervised ones. As such, we propose HyVADRF: Hybrid VADER – Random Forest and Grey Wolf Optimizer Model. Semantic and rule-based VADER was used to calculate polarity scores and classify sentiments, which overcame the weakness of manual labeling, while Random Forest was utilized as its supervised classifier. Furthermore, considering Twitter’s massive size, we collected over 3.6 million tweets and analyzed various dataset sizes as these are related to the model’s learning process. Lastly, Grey Wolf Optimizer parameter tuning was conducted to optimize the classifier’s performance. The results show that 1) HyVADRF Model returned an accuracy of 75.29 %, precision of 70.22%, recall of 87.70%, and F1-score of 78%. 2) The most ideal percentage of dataset size is 90% of the total collected tweets (n=1,249,060). 3) With standard deviations of 0.0008 for accuracy and F1-score and 0.0011 for precision and recall. Hence, HyVADRF Model consistently delivers stable results.

Supervised Sentiment Analysis Algorithms

Sentiment analysis is used to analyse customer sentiment by the process of using naturallanguage processing, text analysis, and statistics. A good customer survey understands thesentiment of their customers—what, how and why they’re saying it. Sentiment dataset can befound mainly in tweets, comments and reviews. Sentiment Analysis understands emotionswith the help of software, and it is playing an inevitable role in today’s workplaces.Sentiment analysis for opinion mining has become an emerging area where more researchand innovations are done. Sentiment or opinion analysis based on a domain is done usingseveral algorithms. Machine learning is a concept among this area. In this, the main focus ison the supervised sentiment analysis or opinion mining algorithms. Supervised learning is adivision coming under machine learning. Different methods of supervised learning andsentiment analysis algorithms are considered and their mode of functioning is studied. Mainfocus of this paper is on the recent trends of research and studies for sentiment classification,taking into consideration the accuracy of different algorithmic techniques that can beimplemented for accurate prediction in sentiment Analysis.

A self structuring artificial intelligence framework for deep emotions modeling and analysis on the social web

The social web has enabled individuals from all walks of life to openly express their emotions and sentiment in relation to current affairs, local issues and personal circumstances. Within the social web, social media encompasses deep emotional expressions that reflect a multitude of personalities and behaviors. Existing research in this space is heavily focused on supervised sentiment analysis and emotion detection, with limited work on modeling these deep emotions, mixed emotions and variations of emotional behaviors from unlabeled and unstructured social media conversations. In this study, we propose a comprehensive framework based on the principles of self-structuring artificial intelligence for emotion modeling and analysis that systematically integrates the modeling capabilities at a granular level on unstructured, unlabeled social media data. The research contributions of this framework are the detection, analysis and synthesis of deep emotion intensity, emotion transitions, emotion latent representations, and profile-based emotion classification.The self-structuring artificial intelligence framework amalgamates an ensemble of novel algorithms to eventuate these contributions. These algorithms extend the current state-of-the-art of natural language processing techniques, word embedding, Markov chains and growing self-organizing maps, specifically for deep emotions modeling and analysis.The framework is empirically evaluated on anonymized conversations from online mental health support forums. The outcomes identify profile-based emotion characteristics, emotion intensities, transitions and an overall latent representation across three distinct mental health groups in these forums. These outcomes are comprehensive in comparison to existing work which singularly focuses on sentiment analysis or emotion detection. The validity and effectiveness of its application on a real-world social media setting further establish the methodological novelty of this ensemble of self-structuring artificial intelligence for deep emotions.

Exploiting semantic relationships for unsupervised expansion of sentiment lexicons

The literature in sentiment analysis has widely assumed that semantic relationships between words cannot be effectively exploited to produce satisfactory sentiment lexicon expansions. This assumption stems from the fact that words considered to be “close” in a semantic space (e.g., word embeddings) may present completely opposite polarities, which might suggest that sentiment information in such spaces is either too faint, or at least not readily exploitable. Our main contribution in this paper is a rigorous and robust challenge to this assumption: by proposing a set of theoretical hypotheses and corroborating them with strong experimental evidence, we demonstrate that semantic relationships can be effectively used for good lexicon expansion. Based on these results, our second contribution is a novel, simple, and yet effective lexicon-expansion strategy based on semantic relationships extracted from word embeddings. This strategy is able to substantially enhance the lexicons, whilst overcoming the major problem of lexicon coverage. We present an extensive experimental evaluation of sentence-level sentiment analysis, comparing our approach to sixteen state-of-the-art (SOTA) lexicon-based and five lexicon expansion methods, over twenty datasets. Results show that in the vast majority of cases our approach outperforms the alternatives, achieving coverage of almost 100% and gains of about 26% against the best baselines. Moreover, our unsupervised approach performed competitively against SOTA supervised sentiment analysis methods, mainly in scenarios with scarce information. Finally, in a cross-dataset comparison, our approach turned out to be as competitive as (i.e., statistically tie with) state-of-the-art supervised solutions such as pre-trained transformers (BERT), even without relying on any training (labeled) data. Indeed in small datasets or in datasets with scarce information (short messages), our solution outperformed the supervised ones by large margins.

Supervised Sentiment Analysis of Science Topics

Twitter is one of the largest sources of real-time information on the Internet and is continuously fed by millions of users around the world. Each of these users publishes text messages with their opinions, concerns, information, or simply their daily happenings. It is a challenge to address the analysis of massive data in the network, just as it is an objective to look for ways to understand everything that data can offer today in terms of knowledge of society and the market. The sector of science communication is still discovering everything that the web 2.0 and social networks can offer to reach all audiences. This article develops a classification model of messages launched on Twitter, on science topics, in Spanish, with machine learning techniques. The training of this type of models requires the creation of a specific corpus in Spanish for the subject of science, which is one of the most laborious tasks. The classifier is able to predict the sentiment of the message in real time on Twitter, with a confidence interval greater than 80%. The results of its evaluation are at 72% accuracy.

An Italian Composite Subjective Well-Being Index: The Voice of Twitter Users from 2012 to 2017

Since 2012, driven by the desire to propose a subjective well-being index (SWBI) complementary to the traditional measures, with high time and space frequency, our team evaluates, analysing Twitter data, a composite index that captures various aspects and dimensions of individual and collective life. The SWBI is a multidimensional indicator whose components were inspired by the dimensions adopted for the Happy Planet Index provided by the New Economic Foundation. In detail, it consists of eight dimensions that describe three different areas: personal well-being, social well-being and well-being at work. The Italian subjective well-being index ($$\hbox {SWBI}_{{ITA}}$$), that we display here, audits the Italian subjective well-being revealed by tweets acquired via the public Twitter API, written in the Italian language, and posted from Italy from January 2012 to December 2017. Around 1–5% of the data includes geo-referenced information, which allows us to provide an index at local level. The Twitter data analysis is carried on with a human supervised sentiment analysis method, the Integrated Sentiment Analysis (iSA) algorithm. In this work, after a weighting procedure adopted to partially overcome the selection bias caused by the use of data from social network, we describe the $$\hbox {SWBI}_{{ITA}}$$ dimensions in the considered period at the regional level. Moreover, for some dimensions, for which a similar currently available measure provided by Italian official statistics exists, comparisons are proposed emphasizing novelties, similarities and differences.

Supervised Sentiment Analysis Model of Textual Content for Images

Sentiment analysis is a domain in machine learning that tries to analyze people’s emotion, feeling, opinion and attitudes towards particular service or product. It aims to extract feelings and opinion from textual reviews; therefore, it is closely related to natural language processing (NLP). Social media has provided a huge amount of text reviews, which is practically impossible to read and analyze the emotions, attitudes and opinions that were expressed in those textual data. Sentiment analysis is a machine learning concept to classify a textual data according to reviewers’ emotion and attitudes about a service or product, which helps in determine strong or weak production. In this paper work we aim to develop a sentiment analysis model of texts for images. Different machine learning algorithms are tested such as Naive Bays, Logistic Regression and Support Vector Machine (SVM), in order to develop a high accuracy sentiment analysis system. The model is developed to determine whether a text has positive or negative emotion for images. The outcome of the project work shows that SVM algorithm has a better performance for such purpose, while Logistic Regression algorithm shows a faster execution time.

Evaluating the performance of sentence level features and domain sensitive features of product reviews on supervised sentiment analysis tasks

With the popularity of e-commerce, posting online product reviews expressing customer’s sentiment or opinion towards products has grown exponentially. Sentiment analysis is a computational method that plays an essential role in automating the extraction of subjective information i.e. customer’s sentiment or opinion from online product reviews. Two approaches commonly used in Sentiment analysis tasks are supervised approaches and lexicon-based approaches. In supervised approaches, Sentiment analysis is seen as a text classification task. The result depends not only on the robustness of the machine learning algorithm but also on the utilized features. Bag-of-word is a common utilized features. As a statistical feature, bag-of-word does not take into account semantic of words. Previous research has indicated the potential of semantic in supervised SA task. To augment the result of sentiment analysis, this paper proposes a method to extract text features named sentence level features (SLF) and domain sensitive features (DSF) which take into account semantic of words in both sentence level and domain level of product reviews. A word sense disambiguation based method was adapted to extract SLF. For every similarity employed in generating SLF, the SentiCircle-based method was enhanced to generate DSF. Results of the experiments indicated that our proposed semantic features i.e. SLF and SLF + DSF favorably increase the performance of supervised sentiment analysis on product reviews.

Proposed Aspect Based Sentiment Analysis system for English reviews

Reviews are a crucial source of opinions that may influence the decision in many areas. So there is a need for an algorithm that is efficient in understanding the aspects that the reviewers have focused on in their reviews and comments on social networks or other web applications. This paper submits a proposed approach for aspect-based sentiment analysis that consists of two steps; the firststep is by a proposedp_chunker algorithm for aspect extraction using Latent Dirchilet Analysis and noun phrase chunking, the second step is sentiment analysis using a proposed hybrid algorithm that depending on both lexicon and supervised sentiment analysis to specify the sentiment for extracted aspects. The proposed paradigm is tested using standard datasets from kaggle for both aspect extraction and sentiment analysis, the result show efficacy in the proposed method.

An efficient preprocessing method for supervised sentiment analysis by converting sentences to numerical vectors: a twitter case study

Along with the significant growth of social media, individuals and companies are increasingly receiving public opinions which direct their decisions. Opinion mining, which is considered as a sub-field of natural language processing, information retrieval, and text mining, is the process of understanding the users’ views from their comment, which have been represented as unstructured texts. Emergence of online social media has led to the production of a huge amount of user comments on websites, and thus, has raised opinion mining as a very useful and challenging problem. In this paper, an efficient preprocessing method for opinion mining is presented and will be used for analyzing users’ comments on Twitter social network. For this purpose, different text preprocessing techniques have been used on the dataset to achieve an acceptable standard text. Word2vec method which is a fast and accurate method have been also exploited to convert the words’ arrays to numerical vectors. Machine learning methods, with supervised learning approach, have been applied on the obtained data after this fast and accurate preprocessing phase. Python and RapidMiner have been used to implement different opinion mining methods and the results of these implementations have been compared and evaluated. The experimental results show that the combined use of the preprocessing method of this paper and support vector machine and artificial neural network have the highest accuracy compared to other methods.

‘Cheap Talk’? Second screening and the irrelevance of TV political debates

This article analyzes 3 months of online debate during the electoral campaign for the 2016 Italian constitutional referendum. Through supervised sentiment analysis, we assess the extent of support for the referendum within the general public of Twitter users ( Twittersphere) by analyzing the voting intentions expressed online in 2,369,333 tweets. Similarly, we exploit the practice of social TV and investigate the support for the referendum expressed by the 160,465 tweets posted by second screeners, that is, the subsample of Twitter users who watched and actively commented on nine political talk shows during the campaign. We compare the mentions and the attitudes of the Twittersphere and the second screeners by means of a lead–lag analysis to test whether the second screeners can act as influencers and trendsetters able to shape or anticipate attention and opinions toward an issue within larger audiences. The results reveal an inverse relationship between the Twittersphere and the second screeners whereby the reactions of the latter diverge from those of the general Twitter public. This finding has implications for the literature on echo chambers and the polarization of social media.

News-based supervised sentiment analysis for prediction of futures buying behaviour

This study examines the predictability of real-time news data on investors’ buying behaviour in the futures market, using supervised sentiment analysis. Market sentiment or traders’ buying behaviour is captured at the bid-ask stage of price formation using the net buying pressure (NBP). Any significant change in NBP patterns defines an “interesting market event”. Real-time news headlines are automatically labelled using interesting market events, assuming a lag between the market information and its impact on buying behaviour. News was found to have an impact on the market buying behaviour of the S&P NIFTY index futures with an optimal lag of 5 minutes. Manual labelling of the news data validated this empirical finding.

Distributed Supervised Sentiment Analysis of Tweets: Integrating Machine Learning and Streaming Analytics for Big Data Challenges in Communication and Audience Research

The large-scale analysis of tweets in real-time using supervised sentiment analysis depicts a unique opportunity for communication and audience research. Bringing together machine learning and streaming analytics approaches in a distributed environment might help scholars to obtain valuable data from Twitter in order to immediately classify messages depending on the context with no restrictions of time or storage, empowering cross-sectional, longitudinal and experimental designs with new inputs. Even when communication and audience researchers begin to use computational methods, most of them remain unfamiliar with distributed technologies to face big data challenges. This paper describes the implementation of parallelized machine learning methods in Apache Spark to predict sentiments in real-time tweets and explains how this process can be scaled up using academic or commercial distributed computing when personal computers do not support computations and storage. We discuss the limitation of these methods and their implications in communication, audience and media studies.El análisis a gran escala de tweets en tiempo real utilizando el análisis de sentimiento supervisado representa una oportunidad única para la investigación de comunicación y audiencias. El poner juntos los enfoques de aprendizaje automático y de analítica en tiempo real en un entorno distribuido puede ayudar a los investigadores a obtener datos valiosos de Twitter con el fin de clasificar de forma inmediata mensajes en función de su contexto, sin restricciones de tiempo o almacenamiento, mejorando los diseños transversales, longitudinales y experimentales con nuevas fuentes de datos. A pesar de que los investigadores de comunicación y audiencias ya han comenzado a utilizar los métodos computacionales en sus rutinas, la mayoría desconocen el uso de las tecnologías de computo distribuido para afrontar retos de dimensión big data. Este artículo describe la implementación de métodos de aprendizaje automático paralelizados en Apache Spark para predecir sentimientos de tweets en tiempo real y explica cómo este proceso puede ser escalado usando computación distribuida tanto comercial como académica, cuando los ordenadores personales son insuficientes para almacenar y analizar los datos. Se discuten las limitaciones de estos métodos y sus implicaciones en los estudios de medios, comunicación y audiencias.

Enfoques big data para la comunicación en salud: análisis de redes y análisis de sentimientos a gran escala

Enfoques big data para la comunicación en salud: análisis de redes y análisis de sentimientos a gran escala

Multimodal learning for topic sentiment analysis in microblogging

Microblogging has become a widely-spread platform of human communication. The massive amount of opinion-rich data in microblogging is helpful to analyze and manage public opinion and social emotion. Different from traditional texts, microblogging data are multimodal, containing multifarious data such as emoticons, image, etc. Most existing sentiment and topic detection approaches treat the unique microblogging data as noise. However, this may lead to unsatisfactoriness in sentiment classification and topic identification. In order to address the issue, we propose a multimodal joint sentiment topic model (MJST) for weakly supervised sentiment analysis in microblogging, which applies latent Dirichlet allocation (LDA) to simultaneously analyze sentiment and topic hidden in messages based the introduction of emoticons and microbloggers personality. Extensive experiments show that MJST outperforms state-of-the-art unsupervised approaches JST, SLDA and DPLDA significantly in terms of sentiment classification accuracy and is promising.