Article
Chemistry, Multidisciplinary
Marco A. Palomino, Farida Aider
Summary: Practical demands and academic challenges have contributed to the thriving research area of sentiment analysis. Pre-processing techniques are needed to clean and normalize the text in social media communications, which often ignore grammar and spelling rules. Despite the extensive discussion on pre-processing in literature, there is no conclusive consensus on the best practices. This study reviews existing research and quantitatively evaluates various combinations of pre-processing components, focusing on Twitter sentiment analysis. The results confirm that the order of pre-processing components significantly affects the performance of naive Bayes classifiers, while lemmatisation improves index performance but not sentiment analysis quality.
APPLIED SCIENCES-BASEL
(2022)
Article
Computer Science, Artificial Intelligence
Marco Pota, Mirko Ventura, Hamido Fujita, Massimo Esposito
Summary: This paper focuses on sentiment analysis of tweets, analyzing the pre-processing of tweets and conducting experiments in different languages to determine the most convenient strategy, aiming to improve the accuracy of sentiment analysis.
EXPERT SYSTEMS WITH APPLICATIONS
(2021)
Review
Chemistry, Multidisciplinary
Yili Wang, Jiaxuan Guo, Chengsheng Yuan, Baozhu Li
Summary: Twitter Sentiment Analysis is an active subfield of text mining, which has attracted considerable interest among researchers. This research provides a comprehensive review of the latest developments in this area, including newly proposed algorithms and applications. The survey classifies each publication based on its significance to specific TSA methods and depicts the current research direction in the field of TSA.
APPLIED SCIENCES-BASEL
(2022)
Article
Computer Science, Artificial Intelligence
Sergio Barreto, Ricardo Moura, Jonnathan Carvalho, Aline Paes, Alexandre Plastino
Summary: With the rapid growth of social media networks, user-generated data on platforms like Twitter has become a valuable resource for decision-making processes. However, the informal and noisy linguistic style of tweets poses challenges for natural language processing tasks, especially sentiment analysis. This study assesses the effectiveness of various neural language models in distinguishing sentiment expressed in tweets.
DATA MINING AND KNOWLEDGE DISCOVERY
(2023)
Article
Multidisciplinary Sciences
Panote Siriaraya, Yihong Zhang, Yukiko Kawai, Peter Jeszenszky, Adam Jatowt
Summary: This paper demonstrates how data from Open Street Map and Twitter can be used to analyze and depict fine-grained human emotions at a city-wide level in San Francisco and London. Through the development of neural network classifiers, emotions are detected from tweets and matched to key locations in Open Street Map. The analysis of the resulting data set reveals the impact of different days, locations, and POI neighborhoods on human emotion expression in the cities.
Article
Computer Science, Artificial Intelligence
Ying Wang, Alvin Wei Ze Chew, Limao Zhang
Summary: This study develops a deep learning framework to quantify public sentiments towards COVID-19 and uses these sentiments to forecast the daily growth rate of confirmed COVID-19 cases globally. The results show that the global community evokes both positive and negative sentiments towards COVID-19 over time.
APPLIED SOFT COMPUTING
(2022)
Article
Computer Science, Artificial Intelligence
Sarah Alhumoud, Asma Al Wazrah, Laila Alhussain, Lama Alrushud, Atheer Aldosari, Reema Nasser Altammami, Njood Almukirsh, Hind Alharbi, Wejdan Alshahrani
Summary: This article explores Arabic Sentiment Analysis for Vaccine-Related COVID-19 Tweets (ASAVACT) and presents the largest Arabic Twitter corpus on COVID-19 vaccination. State-of-the-art deep learning models are utilized, with the ensemble model showing the highest accuracy improvement.
PEERJ COMPUTER SCIENCE
(2023)
Article
Computer Science, Information Systems
Roberto Casarin, Juan C. Correa, Jorge E. Camargo, Silvana Dakduk, Enrique ter Horst, German Molina
Summary: This article proposes the use of computationally efficient inverse regression Bayesian method for analyzing the propagation of political messages on Twitter. It highlights how politicians can identify sensitive word combinations to increase the probability of message retweets, impacting political outcomes. The method enables adjustment of political messages to increase voter engagement in political campaigns.
JOURNAL OF INFORMATION SCIENCE
(2021)
Article
Computer Science, Information Systems
Hassan Nazeer Chaudhry, Yasir Javed, Farzana Kulsoom, Zahid Mehmood, Zafar Iqbal Khan, Umar Shoaib, Sadaf Hussain Janjua
Summary: This study conducted sentiment analysis on Twitter to compare public opinions on the 2016 and 2020 U.S. elections, finding that election outcomes generally align with sentiments expressed on social media. The research also analyzed outliers and controversial swing states, validating election results against sentiments expressed online.
Article
Chemistry, Multidisciplinary
Chung-Hong Lee, Hsin-Chang Yang, Yenming J. Chen, Yung-Lin Chuang
Summary: The emerging application field of using Twitter messages and algorithmic computation to detect real-time world events has become a new paradigm in data science applications. The study focuses on integrating real-time event monitoring and intelligence gathering functions to provide updated event summaries, combined with pre-trained language models for summarizing emergent events.
APPLIED SCIENCES-BASEL
(2021)
Article
Computer Science, Information Systems
Priyavrat Chauhan, Nonita Sharma, Geeta Sikka
Summary: The paper emphasizes the importance of data pre-processing in sentiment analysis of Twitter data. It provides detailed analysis and methods for understanding and handling Twitter data during elections. The study argues that personal tweets from users are more accurate in assessing public opinion than tweets from news or media sources. It also highlights the significant role of emojis, punctuations, stopwords, emphasized words, and specific regions within tweets in sentiment analysis. The paper presents a novel set of pre-processing steps to filter and clean tweets without losing vital information. Experimental results reveal the impact of media tweets and specific regions on sentiment analysis and the need for accurate data selection. Exploratory analysis further examines public sentiments towards political terms inferred using top2vec topic modeling.
MULTIMEDIA TOOLS AND APPLICATIONS
(2023)
Article
Computer Science, Artificial Intelligence
Zijian Feng, Hanzhang Zhou, Zixiao Zhu, Kezhi Mao
Summary: TTA is a novel data augmentation technique for sentiment analysis that improves the model's generalization capability through probabilistic word sampling for synonym replacement and application of zero masking or contextual replacement to discriminative words irrelevant to sentiment.
EXPERT SYSTEMS WITH APPLICATIONS
(2022)
Article
Computer Science, Artificial Intelligence
Monali Bordoloi, Saroj Kumar Biswas
Summary: Sentiment analysis is a solution for extracting summarized opinions or details from a large data source. A thorough analysis of the process for developing an efficient sentiment analysis model is desired. Factors such as word extraction, sentiment classification, dataset, and data cleansing greatly affect the performance of a sentiment analysis model. This paper provides in-depth knowledge of different techniques, algorithms, and factors associated with designing effective sentiment analysis models. It also critically assesses different modules of a sentiment analysis framework, discusses shortcomings of existing methods, and proposes potential multidisciplinary applications and research directions.
ARTIFICIAL INTELLIGENCE REVIEW
(2023)
Article
Multidisciplinary Sciences
Khizar Qureshi, Tauhid Zaman
Summary: In this study, a new model is proposed to predict the future performance of cryptocurrencies using social media data. The engagement coefficients of cryptocurrencies are found to be correlated with their future returns. Cryptocurrencies with extreme engagement coefficients tend to have lower returns, indicating either a lack of interest or the presence of artificial activity from bots. Simple investment strategies based on selecting cryptocurrencies with high engagement coefficients outperform in holding periods of a few months.
Article
Computer Science, Theory & Methods
Alhanouf Alduailej, Abdulrahman Alothaim
Summary: This paper introduces AraXLNet, an Arabic language model based on XLNet, and applies it to sentiment analysis tasks in Arabic. The results demonstrate the superior performance of AraXLNet over the previous AraBERT model on multiple benchmark datasets.
JOURNAL OF BIG DATA
(2022)