A review on sentiment analysis and emotion detection from text

As a result of the rapid development of the Internet era and the increasing prevalence of Internet use, social networking platforms have become an essential means of exchanging feelings with people worldwide. The ability to submit real-time messages representing their opinions on various subjects, talk about common problems or complaints, and express their feelings about anything that matters is available to web users. Social media networks generate a substantial amount of unstructured data on the Internet each and every second. Many people are interested in understanding human psychology because of the abundance of data. Sentiment analysis is used, which determines if the person may have negative, positive, or neutral thoughts on an incident. Emotion detection, which determines a person’s emotional state, is necessary in particular situations where the use of sentiment analysis is insufficient. Deep learning has become the most popular sentiment analysis and emotion recognition technique. However, the need for many tagged data to achieve better performances demonstrates a more effective strategy to overcome this restriction. An emerging topic, "Meta-learning", seeks to enhance algorithms in varied ways, which include generalizability and data efficiency. This review study offers insight into multiple sentiment analysis levels, diverse emotion models, fundamental meta-learning ideas, an analysis of sentiment analysis and emotion detection techniques and future directions.

Due to the Internet's rapid expansion, social networking sites have emerged as crucial tools for sharing emotions with the entire globe. Many people share their thoughts or opinions through text, images, audio, and video On the other hand, text communication over web-based network media can be a little overwhelming. Social media web sites generate a substantial volume of unstructured data on the Internet every second. It is essential to analyze data as soon as it is generated in order to understand human psychology. Sentiment analysis, which detects polarity in texts, can help with this. It evaluates the author's attitude toward a particular thing, administration, person, or place and determines if that attitude is negative, positive, or neutral. A person's emotional/mental condition can be reliably detected using emotion detection. This article describes different levels of sentiments, emotions models, and techniques for detecting sentiments from text. The study also emphasis the challenges associated with assessing mood and emotion. Last but not least, this paper also highlights the difficulties encountered during sentiment and emotion analysis.

El objetivo principal de esta tesis doctoral es mejorar el análisis de sentimientos y emociones de texto en redes sociales, aunando técnicas de procesamiento de lenguaje natural, datos enlazados y análisis de redes sociales. La investigación se divide en tres partes muy diferenciadas. Primero, se desarrolló un vocabulario semántico para describir emociones y procesos de análisis de sentimientos, alineado con la ontología de “procedencia” PROV-O. Este vocabulario permite seguir un enfoque de datos enlazados en el análisis de emociones, tanto en la anotación de recursos (datasets y lexicons), como en la publicación de servicios semánticos de análisis de emociones. Asimismo, se extendió el vocabulario de referencia para opiniones, Marl, para alinearlo con Prov-O. En segundo lugar, se han modelado los diferentes componentes de los servicios de análisis de sentimientos y emociones, así como los requisitos para crear servicios abiertos, interoperables y que se puedan combinar para lograr análisis avanzados. El resultado es un marco de desarrollo y modelado de servicios, enfocado en la modularidad. Además, se ha desarrollado una implementación de referencia que permite a crear y publicar servicios de análisis de sentimientos y emociones. En tercer lugar, se ha caracterizado el contexto social, que es el conjunto de información en una red social que complementa al mensaje, y que puede ser utilizado para mejorar el análisis de sentimientos del mensaje. También se ha desarrollado una taxonomía de enfoques de análisis de sentimientos basada en la forma en que el contexto social es construido y utilizado en el análisis. Seguidamente, se han investigado modelos de análisis de sentimientos que utilizan contexto social enriquecido mediante análisis de redes sociales. Por último, para explorar el potencial de las diferentes teorías sociales para el análisis de sentimientos se ha desarrollado una plataforma de simulación social, en la que se han implementado varios modelos de propagación de rumores y emociones. ----------ABSTRACT---------- The main goal of this thesis is to improve sentiment and emotion analysis of text in social media through a combination of natural language processing, linked data and social network analysis. To achieve this goal, we have divided our research into three parts. First, we developed a semantic vocabulary to describe emotions, emotion models and emotion analysis activities. This vocabulary enables a linked data approach to emotion analysis, including in the annotation and processing of resources (e.g., datasets and lexicons), and the development of public semantic emotion analysis services. We also extended the most popular vocabulary for opinions and sentiment, Marl, to include concepts of sentiment analysis activities. Secondly, we modeled the different components in a sentiment or emotion analysis service, as well as the requirements to create public and interoperable services that can be composed to produce advanced analyses. The result is a framework to model and develop modular services. We also developed a reference implementation of this framework, which can be used by researchers and developers to create and publish new sentiment and emotion analysis services. Thirdly, we studied and formalized the concept of social context, which is the information in a social network that accompanies a text message and can be used to improve the analysis of said text. We also developed a taxonomy of approaches to sentiment analysis based on how they gather social context and how they exploit it in the analysis. In addition to characterizing social context, we investigated several models of sentiment analysis that enrich social context through social network analysis. Lastly, we developed a social simulation platform, in which we modelled several rumor and emotion propagation behaviors.

The health sector has benefited in many ways from the data science and artificial intelligence (AI). One of the most promising applications of data science and AI in the healthcare field is sentiment analysis and emotion detection. Sentiment analysis is the automatic categorization of sentiment in a free text, whereas emotional detection categorizes emotion on a human face using a sophisticated image dispensation. This chapter aimed to focus on Sentiment Analysis and Emotion detection applications related to wellbeing healthcare through a systematic review of the recent literature. With the support of AI methods and other mathematical models, sentiment analysis can offer significant assistance to healthcare professionals, especially psychiatrists to understand the mental health and psychological problems of wellbeing. In general, people with certain intolerable problems, serious illnesses, addictions to something, suicide victims, and caregivers use social networks, health websites, and other web portals to share their sentiments. These are important data sources for sentiment analysis related to health. Emotion detection and recognition mechanisms use facial expressions for emotions such as joy, sadness, surprise, and anger, and in addition, capture “micro-expressions” or controlled expression of body language as the main source of data. Analysis outcomes help health professionals to decide when patients need help or need medication. In conclusion, health professionals and community service volunteers or caregivers can use the results of the sentiment and emotion detection analysis to help with wellbeing when they need it. The accuracy of the analysis results can be improved by combining the analysis of human expressions from a variety of forms such as texts, facial expressions, body language, and speech.KeywordsSentiment analysisEmotion detectionAnd recognitionHealthcareAI applicationsData science applications

Abstract(NLP) is an acronym for natural language processing. It refers to the branch of computer science—and more specifically, the branch of artificial intelligence concerned with giving computers the ability to understand the text and spoken words in the similar way human beings can do. Emotion detection (ED) is a sentiment analysis branch dedicated to emotional extraction and analysis. Speech, facial expressions, and body language may determine human emotions efficiently. Emotion detection is an emerging field of research in sentiment analysis for effective interaction between humans and computers. Emotion detection and text recognition is a recent study subject that is strongly connected with sentiment analysis. Positive, neutral, or negative sensations are identified in a text, whereas the objective of the emotional analysis is to discover and recognize different kinds of feelings through the use of words such as wrath, disgust, fear, happiness, sorrow, and surprise. A multitude of research in the field of text mining and analysis is being conducted due to the convenience of data acquisition and the enormous benefits offered by it. This article examines the concept of emotion detection (ED) from text, image and outlines the primary approaches taken in the construction of text and image-based emotion detection (ED) systems by researchers. This article further discusses about how we are going to use sentiment analysis in emotion detection to analyze what exactly is the intent of the sender toward the receiver, how we are going to apply emotion detection (ED) to bifurcate the various emotions of the sender. This research paper proposes the discussion related to the handouts, findings, datasets used, and the combined outcome obtained after analyzing the dataset. This research is based on two (ED) emotion detection techniques: I) Text based (ED) II) Image based (ED). Researcher aims to analyze a plain text and images on any platform, be it reviews, comments or image on a blog post or social media applications, etc. It further focuses on detecting the emotion of the sender or receiver by using sentiment analysis.( Sentiment Analysis: The Go-To Guide monkeylearn.com), Sentiment Analysis.KeywordsNatural language processingSentiment analysisEmotion detectionFace recognitionImage processingComputer vision

Nowadays, with the increasing number and use of social media platforms, people now share their experiences about a product they have bought or a place they have been to on social media platforms more frequently. Considering the volume of data on social media platforms, it is considered that there is some meaningful information for institutions or companies in the reviews and experiences shared on social media platforms. As such, it is important to improve the methods of extracting meaningful information from the reviews and experiences shared on social media and to know which method is better. In this study, the classification successes of the bag of words and the fastText word representation methods, which are among the word representation methods in sentiment analysis methods mentioned above, were compared by using Turkish reviews performed for touristic places. Besides, while performing the comparison process, it was measured whether the process of separating the words into their roots and negation of the words, which is the preliminary stage of the sentiment analysis process, contributed to the classification success. In the study, both two-class (positive, negative) sentiment analysis and three-class (positive, negative, neutral) sentiment analysis were performed. Six data sets were created to carry out the mentioned comparison operations. The data sets were first classified using the Naive Bayes (NB), Multinomial Naive Bayes (MNB), k-Nearest Neighbor (k-NN) and Support Vector Machines (SVM) algorithms, which are frequently used in text mining, and based on bag of words word representation method, they were classified with WEKA program. After the test results of all data sets were obtained according to the bag of words word representation method, the tests of the fastText word representation method were carried out using the fastText library of the Python programming language. Classification procedures were carried out with 10-fold cross-validation methods, and f-score values of the classification processes were obtained. Finally, it was determined that bag of words word representation method performed a more successful classification than the fastText word representation method in two-class emotion analysis, while the fastText word representation method performed a more successful classification process than bag of words word representation method in three-class emotional analysis. It was observed that the process of separating the words into their roots and negating the words, which are the preliminary processes of sentiment analysis, did not contribute positively or negatively to the classification processes performed with the fastText word representation method. However, it was determined that it had a minor contribution to sentiment analysis processes performed by using bag of words word representation method. In the two-class sentiment analysis, the most successful classification result was achieved by using the machine learning model created with the SVM algorithm with the value of 0.91 f-score employing bag of words word representation method. In the three-class sentiment analysis, the most successful classification result was achieved with the machine learning model created using the fastText word representation method with the value of 0.78 f-score.

Text generation approaches in health care communication have evolved along 2 major paths. The first path involves generative adversarial networks, progressing from basic architectures to specialized variants like Text-to-Text Generative Adversarial Network (TT-GAN) and Time and Frequency Domain-Based Generative Adversarial Network (TF-GAN), which address challenges in discrete text generation through techniques such as Gumbel-Softmax and reinforcement learning. The second path emerges from transformer-based architectures, particularly Generative Pretrained Transformer-2 (GPT-2), which uses extensive pretraining and self-attention mechanisms to generate contextually appropriate text. GPT-2's transformer architecture enhances persuasive health communication by generating personalized messages using various strategies like task support, dialogue support, and social support for effective health interventions. This study aimed to use GPT-2 as a generative method to construct persuasive text in a dataset and compare the performance of sentiment analysis and emotion detection analysis. We combined sentiment analysis tools (VADER [Valence Aware Dictionary and Sentiment Reasoner] and TextBlob) with emotion detection methods (Text2Emotion and NRCLex [National Research Council Lexicon]) to analyze health coaching messages across different persuasive types: reminder, reward, suggestion, and praise. TextBlob and VADER achieved accuracies of 57% and 69%, respectively, while RoBERTa (robustly optimized BERT approach)-sentiment outperformed them with an accuracy of 88%. Emotion detection showed a high prevalence of "joy" and "happy" labels (93.69% positive skew). While transformers excel in accuracy, lexicon-based models like VADER offer a better performance-efficiency balance for real-time health communication systems. For emotion detection, all categories showed perfect accuracy (1.0), while trust showed mixed results, with precision, recall, and F1-score values ranging from 0.81 to 0.96. The emotion detection analysis revealed varying success rates across different emotions, with some categories, such as anger and neutral, showing reasonable performance and others, such as trust, showing mixed performance. This research contributes to understanding the emotional dynamics of persuasive health communication and highlights both the capabilities and limitations of current natural language processing tools in analyzing health-related persuasive messaging. This proof-of-concept study using synthetically generated data establishes a methodological framework for multimodal sentiment and emotion analysis. The findings require validation with real-world health coaching messages before clinical deployment.

With the spread of social media services in Arabic societies, it leads to the explosive growth of Arabic posts, or comments. These services generate a huge volume of opinionated data on different topics such as politics and businesses. Analyzing valuable subjective information from data would assist in a better understanding and making decisions. Therefore, sentiment analysis coincides with social media networks and has become the most interesting research field in the sentiment analysis process. However, there are several challenges faced the sentiment analysis process. Arabic Sentiment analysis is indeed in its infantile stage and it has not obtained thoroughly attention wherein several challenges still need to address. Some of these challenges result from the complexity of Arabic natural language and other challenges result from social media platform itself. In this manuscript, we first study the impact of social media challenges on the challenges of Arabic language. Our findings show that such challenges add more complexities to the sentiment analysis process. Based on these findings, we review the contributed proposals, which give rise on analyzing Arabic social media data. Our review methodology is based on a set of criteria, which we propose to assess the advantages and limitations of these proposals. The interesting point here is to help researchers identify the social sentiment analysis problems along with a comprehensive survey on the sentiment analysis levels and classification approaches. Finally, we compare these proposals in terms of the average accuracy and suggest a new hybrid approach based on our findings.

Sentiment and emotion analysis on social media is an interesting study because it reveals the emotional state of the public in a domain. The challenges in sentiment analysis research in Indonesian are inefficient preprocessing, inaccurate feature extraction methods, and low classification accuracy by machine learning. One aspect of sentiment analysis is fanaticism. Fanaticism contains an emotional element in sentiment analysis. This article discusses how to detect opinions that contain political fanaticism, then categorize them into several polarities of political fanaticism. Feature extraction is done by processing sentiment, anger, happiness, disgust, surprise, fear, and hate speech analysis. Knowledge for classification is K-NN, Naive Bayes, Random Forest, and Decision Tree. The aim is to find out the best combination of machine learning methods for feature extraction and finally used for fanaticism categorization. The best method is Random Forest with an accuracy of 81% and will be used as a final method for monitoring fanaticism on social media.

With the advent of technology, most medical organizations have developed medical platforms where a large group of patients computes the textual data. Depression is a common type of mental disorder that has a relative impact on society. People use social media platforms and share their emotions, ideas, and thoughts with others. Hence, an automated health monitoring scheme is essential to monitor the health status of the patients. Through the monitoring of the social media platforms, the medical sentiments of the persons can be analyzed by the user comments. This paper presents the standard domains for analyzing the medical sentiments emphasized on depression. The process of sentimental analysis is described with its steps. The procedure comprises the collection of medical data from social sites, preprocessing, extracting the features and applying a classifier to classify the data. There are several existing methods of sentimental analysis based on the traditional machine learning methods, semi-supervised statistical methods, deep learning algorithms like long short-term memory classification model and many more. In addition, the notion of medical sentiment analysis has been described. In this paper, we discussed the existing methods for examining medical emotions using social media platforms such as Facebook and Twitter. The process of medical sentiment analysis is depicted using the approaches provided, which are then compared using performance metrics to identify its applications and challenges.KeywordsDepressionMedical sentiment analysisSocial sitesMachine learning classifiersHealth monitoring scheme

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

Social networking platforms have witnessed tremendous growth of textual, visual, audio, and mix-mode contents for expressing the views or opinions. Henceforth, Sentiment Analysis (SA) and Emotion Detection (ED) of various social networking posts, blogs, and conversation are very useful and informative for mining the right opinions on different issues, entities, or aspects. The various statistical and probabilistic models based on lexical and machine learning approaches have been employed for these tasks. The emphasis was given to the improvement in the contemporary tools, techniques, models, and approaches, are reflected in majority of the literature. With the recent developments in deep neural networks, various deep learning models are being heavily experimented for the accuracy enhancement in the aforementioned tasks. Recurrent Neural Network (RNN) and its architectural variants such as Long Short Term Memory (LSTM) and Gated Recurrent Unit (GRU) comprise an important category of deep neural networks, basically adapted for features extraction in the temporal and sequential inputs. Input to SA and related tasks may be visual, textual, audio, or any combination of these, consisting of an inherent sequentially, we critically investigate the role of sequential deep neural networks in sentiment analysis of multimodal data. Specifically, we present an extensive review over the applicability, challenges, issues, and approaches for textual, visual, and multimodal SA using RNN and its architectural variants.

Chapter 4 - Linked Data Models for Sentiment and Emotion Analysis in Social Networks

Globally, social media is gaining popularity and redefining how people interact with one another online. Malaysian individuals, for example, are increasingly reliant on social media platforms such as Facebook and Twitter as well as LinkedIn, Pinterest, Instagram, and other similar sites. Consider sentiment analysis to be a sub-category of social listening. A social media sentiment analysis has uncovered the public's current feelings on a particular topic or brand. Sentiment analysis is a technique for characterizing and capturing emotional states from unstructured text. The most important part of sentiment analysis is to evaluate a body of text to comprehend the opinion expressed by it. It usually assigns a polarity of "positive", "negative" or "neutral". It uses an algorithmic technique to capture people's thoughts, sentiments, and emotions by incorporating Natural Language Processing and Machine Learning technology. Sentiment analysis in Malaysia's social media is challenging to perform since posts are frequently written in a mixed language, usage of English and Malay with embedded jargon and various district dialect. The classification was performed based on Malaysia halal certification scheme for each tweet to acquire the class label's frequency value based on the sentiment analysis process's polarity results. It will demonstrate social media users' proclivity for posting and can act as a reference point for users when making decisions. An analysis of amounted 500 tweets with the hashtag #sijilhalal elicited information regarding people's feelings, preconceptions, and attitudes toward various issues related to halal certification in Malaysia. The discovery of a person's emotions concerning halal topics is visualized. Muslims' views are of importance to #sijilhalal awareness.

Sentiment analysis or opinion mining has become one of the fastest growing areas now. Though the journey started since 1990s but huge outbreak of sentiment analysis occurred after 2004. With the increasing pressure of new era, new technology, more complex and busy lifestyle, mental health issues are also becoming more serious concerns. In this survey paper, a brief evolution history of sentiment analysis has been discussed. Briefly, emotion detection through facial expression has also been addressed. Commonly used approaches and technology used for sentiment analysis and emotion detection has been discussed with the comparison of available technologies. With the methodology used for sentiment analysis, an insight view of mental health concerns where sentiment analysis can play a vital role, has been discussed. Nowadays, our youth generation is using social website in a large scale to express their mental status, as a way of entertainment, to express their general opinion regarding any topic or issue. Hence the large web data now has gained the ability to show the overall mental condition for a large community. After corona outbreak mental issues have been increased significantly as well as use of social media also has been increased incredibly due to lockdown and work from home lifestyle. So by using the vast data of web platform, we can analyze the recent situation of mental health issues and also predict near future concerns. Although in accuracy of sentiment analysis, we are still facing many challenges with our existing algorithms, but there are a lot of future scope in this field.