Microblog Messages Research Articles

Initial exploration into sarcasm and irony through machine translation

In this paper, we investigate sarcasm and irony as seen through a novel perspective of machine translation. We employ various techniques for translation, comparing both manually and automatically translated datasets of irony and sarcasm. We first clarify the definitions of irony and sarcasm and present an exhaustive field review of studies on irony both from purely linguistic as well as computational linguistic perspectives. We also propose a novel evaluation metric for the purpose of evaluating translations of figurative language, with a focus on machine-translated irony and sarcasm. The constructed English and Chinese parallel dataset includes polarized content from tweets as well as forum posts, categorized by irony types. The preferred translation model, mBART-50, is identified through a thorough experimental process. Optimal translation settings and the best-finetuned model for irony are explored, with the most effective model being finetuned on both ironic and non-ironic data. We also experimented which types of irony are best suitable for training in this specific task — short microblogging messages or longer forum posts. Moreover, we compare the capabilities of a well fine-tuned mBART to a prompt-based method using the recently popular ChatGPT model, with the conclusion that the former still outperforms the latter, although ChatGPT without any training can be considered as a “good enough” ad hoc solution in the case of a lack of data for training. Finally, we verify if the translated data – either manually, or with an MT model – can be used as training data in a task of irony detection. We believe that the presented research can be expanded into languages other than the presented here Chinese and English, which together with the ability to detect various categories of irony, could contribute to deepening the understanding of figurative language, especially irony and sarcasm.

Analysis of Guangzhou city image perception based on weibo text data (2019–2023)

With the popularization of smart mobile terminals and social media, a large amount of data containing textual information about the city has been generated on social media platforms, covering all areas of the city. This provides a new way for the study of comprehensive perception of city image. In the Internet era, users express their opinions about cities through social media platforms (e.g., Sina Weibo), and mining this information helps to understand the image of cities on mainstream social media and to target positive images to improve the competitiveness of the city's image. In this paper, 370,000 microblog messages related to “Guangzhou City” between 2019 and 2023 are collected using web crawler technology, and three typical text analysis methods are adopted: Term Frequency-Inverse Document Frequency (TF-IDF), Latent Dirichlet Allocation (LDA), and Sentiment Analysis (SnowNLP), to understand the characteristics of Guangzhou city image. gain an in-depth understanding of Guangzhou's urban image characteristics. The study shows that extensive data analysis methods based on text mining can perceive the dynamics and trends of the city in a timely manner, refine the characteristics of Guangzhou's urban image, and propose communication strategies for Guangzhou's image. This study aims to mine Guangzhou's urban image presented on Weibo, provide data support for relevant departments in China and Guangzhou to formulate communication strategies, and provide references for other cities to manage their urban image.

Twitter users perceptions of AI-based e-learning technologies

Today, teaching and learning paths increasingly intersect with technologies powered by emerging artificial intelligence (AI).This work analyses public opinions and sentiments about AI applications that affect e-learning, such as ChatGPT, virtual and augmented reality, microlearning, mobile learning, adaptive learning, and gamification. The way people perceive technologies fuelled by artificial intelligence can be tracked in real time in microblog messages promptly shared by Twitter users, who currently constitute a large and ever-increasing number of individuals. The observation period was from November 30, 2022, the date on which ChatGPT was launched, to March 31, 2023. A two-step sentiment analysis was performed on the collected English-language tweets to determine the overall sentiments and emotions. A latent Dirichlet allocation model was built to identify commonly discussed topics in tweets. The results show that the majority of opinions are positive. Among the eight emotions of the Syuzhet package, ‘trust’ and ‘joy’ are the most common positive emotions observed in the tweets, while ‘fear’ is the most common negative emotion. Among the most discussed topics with a negative outlook, two particular aspects of fear are identified: an ‘apocalyptic-fear’ that artificial intelligence could lead the end of humankind, and a fear for the ‘future of artistic and intellectual jobs’ as AI could not only destroy human art and creativity but also make the individual contributions of students and researchers not assessable. On the other hand, among the topics with a positive outlook, trust and hope in AI tools for improving efficiency in jobs and the educational world are identified. Overall, the results suggest that AI will play a significant role in the future of the world and education, but it is important to consider the potential ethical and social implications of this technology. By leveraging the positive aspects of AI while addressing these concerns, the education system can unlock the full potential of this emerging technology and provide a better learning experience for students.

Tweet topics and sentiments relating to distance learning among Italian Twitter users

The outbreak of COVID-19 forced a dramatic shift in education, from in-person learning to an increased use of distance learning over the past 2 years. Opinions and sentiments regarding this switch from traditional to remote classes can be tracked in real time in microblog messages promptly shared by Twitter users, who constitute a large and ever-increasing number of individuals today. Given this framework, the present study aims to investigate sentiments and topics related to distance learning in Italy from March 2020 to November 2021. A two-step sentiment analysis was performed using the VADER model and the syuzhet package to understand the overall sentiments and emotions. A dynamic latent Dirichlet allocation model (DLDA) was built to identify commonly discussed topics in tweets and their evolution over time. The results show a modest majority of negative opinions, which shifted over time until the trend reversed. Among the eight emotions of the syuzhet package, ‘trust’ was the most positive emotion observed in the tweets, while ‘fear’ and ‘sadness’ were the top negative emotions. Our analysis also identified three topics: (1) requests for support measures for distance learning, (2) concerns about distance learning and its application, and (3) anxiety about the government decrees introducing the red zones and the corresponding restrictions. People’s attitudes changed over time. The concerns about distance learning and its future applications (topic 2) gained importance in the latter stages of 2021, while the first and third topics, which were ranked highly at first, started a steep descent in the last part of the period. The results indicate that even if current distance learning ends, the Italian people are concerned that any new emergency will bring distance learning back into use again.

Knowledge Graph Construction for Intelligent Media Based on Mobile Internet

Mobile Internet-based intelligent media has become a popular academic topic. This study uses the CiteSpace visualisation tool and Scientific Citation Index Expanded database to comb the existing research in the field of intelligent media from a quantitative perspective. A total of 7248 English papers were published on the topic of “intelligent media” from 2012 to 2021, and 145 highly cited papers refined were analysed. Scientific knowledge graphs were analysed from six dimensions: annual publication quantity, country of publication, institution of publication, author, keywords, and cited references. In the last 10 years, the research literature on intelligent media has been found to increase annually. Presently, the People’s Republic of China and the United States of America have a high proportion of documents in this field. Chinese universities and institutions have achieved significantly in terms of the quantity and quality of documents. From the perspective of the whole intelligent media discipline, the high-yield author group has not been formed, and there is minimal cooperation amongst authors. Popular intelligent media topics include film, social media, machine learning, swarming motility, data mining, and artificial intelligence. Subject words of the main research directions are event recognition, fake news, Cable News Network model, reconfigurable intelligent surface, comprehensive survey, microblog message, strain sensor, and traffic event. Combined with popular topics and time zone maps, the future research frontier in the field of smart media is identified.

A Bootstrapping Approach to Identifying Relevant Tweets for Social TV

Manufacturers of TV sets have recently started adding social media features to their products. Some of these products display microblogging messages relevant to the TV show which the user is currently watching. However, such systems suffer from low precision and recall when they use the title of the show to search for relevant messages. Titles of some popular shows such as Lost or Survivor are highly ambiguous, resulting in messages unrelated to the show. Thus, there is a need to develop filtering algorithms that can achieve both high precision and recall. Filtering microblogging messages for Social TV poses several challenges, including lack of training data, lack of proper grammar and capitalization, lack of context due to text sparsity, etc. We describe a bootstrapping algorithm which uses a small manually labeled dataset, a large dataset of unlabeled messages, and some domain knowledge to derive a high precision classifier that can successfully filter microblogging messages which discuss television shows. The classifier is designed to generalize to TV shows which were not part of the training set. The algorithm achieves high precision on our two test datasets and successfully generalizes to unseen television shows. Furthermore, it compares favorably to a text classifier specifically trained on the television shows used for testing.

A Penny for Your Tweets: Campaign Contributions and Capitol Hill Microblogs

Who influences a politician’s public statements? In this paper, we explore one plausible explanation: that financial incentives from campaign contributors affect what politicians say. Based on this idea, we design text-driven models for campaign contribution profile prediction. Using a large corpus of public microblog messages by members of the U.S. Congress, we find evidence for such an association, at the level of contributing industries. We find complementary strengths in a simple model (which has better predictive accuracy) and a more complex model (which gives a more intuitive, human-interpretable explanation).

Use of Microblogging, Social Networking, and Short Messages in E-learning for Information Culture Building

This work examines the effect of microblogging, social networking, and short messaging in e-learning on the process of information culture formation. The study takes an experimental training course "The Psychology of Management" designed with Moodle, which implies the use of Instagram, and applies it to a sample of master’s level students. Learners were recruited from the Higher School of Economics and Business (Al-Farabi Kazakh National University) together with the Institute of Economics and Management in Agribusiness (Russian State Agrarian University - Timiryazev Moscow Agricultural Academy (RSAU – MTAA)). Students were asked to briefly express their thoughts, attitudes, and views toward the given topic on Instagram by creating blog posts and leave motivational and constructive comments under the blogs of other group members. The emotional responses of learners to the innovative style of teaching (i.e., the use of blogs) and its effect on the learning process were assessed using surveys. The study discusses the effect of blogging on Instagram as an educational tool. The majority of respondents reported a positive effect of blogs. According to learners, the use of Instagram gave the learning journey a makeover, making it relevant, effortless, and consistent. The use of blogs was found to create favorable conditions for collaborative learning, creative self-expression, and systematization of knowledge and experiences. Students reported that doing blogs was entertaining, motivating, and beneficial to their critical thinking, writing skills, and communication competency.

Microblog Topic-Words Detection Model for Earthquake Emergency Responses Based on Information Classification Hierarchy

Social media data are constantly updated, numerous, and characteristically prominent. To quickly extract the needed information from the data to address earthquake emergencies, a topic-words detection model of earthquake emergency microblog messages is studied. First, a case analysis method is used to analyze microblog information after earthquake events. An earthquake emergency information classification hierarchy is constructed based on public demand. Then, subject sets of different granularities of earthquake emergency information classification are generated through the classification hierarchy. A detection model of new topic-words is studied to improve and perfect the sets of topic-words. Furthermore, the validity, timeliness, and completeness of the topic-words detection model are verified using 2201 messages obtained after the 2014 Ludian earthquake. The results show that the information acquisition time of the model is short. The validity of the whole set is 96.96%, and the average and maximum validity of single words are 78% and 100%, respectively. In the Ludian and Jiuzhaigou earthquake cases, new topic-words added to different earthquakes only reach single digits in validity. Therefore, the experiments show that the proposed model can quickly obtain effective and pertinent information after an earthquake, and the complete performance of the earthquake emergency information classification hierarchy can meet the needs of other earthquake emergencies.

To Entertain or to Serve: Chinese and US Banks’ Online Identity Based on a Genre Analysis of Social Media

<bold xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">Background:</b> Social media, increasingly conceptualized as communicative genres, has become a tool for individuals to promote social bonds, as well as a platform for corporations to engage with the public and build corporate identity. <bold xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">Literature review:</b> Genre analysis has seldom been used in investigating social media use and corporate identity construction, especially the relationship between corporate identity and the interaction of message content and communicative style. Therefore, this study intends to implement Lomborg's framework of genre analysis for social media studies to compare Weibo posts of Chinese banks with Twitter tweets of US banks in terms of thematic orientations (message content) and communicative styles (indicated by speech act use) and thus examine their corporate identities. <bold xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">Research questions:</b> 1. What thematic orientations do Chinese and American banks develop in their microblogging messages? 2. What communicative styles do Chinese and American banks develop in their microblogging messages? 3. How are corporate identities formed through the interaction of thematic orientations and communicative styles of the microblogging messages of Chinese and American banks? <bold xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">Methodology:</b> We collected posts and tweets from the official microblogging accounts of 10 Chinese and US banks on the Fortune Global 500 list, identified thematic orientations and communicative styles, and conducted a comparative analysis of corporate identities based on the interaction of thematic orientations and communicative styles. <bold xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">Results and conclusions:</b> In analysis of thematic orientations, Chinese banks are more likely to post messages of <sc xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">entertainment, operation,</small> and <sc xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">product</small> . In terms of communicative styles, their messages are more likely to be characterized by directness, inexpressiveness, and independence politeness overall. Specifically, they tend to send <sc xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">entertainment</small> posts directly and in the involvement politeness style, <sc xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">operation</small> posts indirectly and in the independence politeness style, and <sc xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">product</small> posts in the independence politeness style. As such, Chinese banks may best be described as the audience's friendly companions, objective and authoritative press spokesmen, and competitive innovators, a combination that implies values of relationship, authority, and competition in collectivism-oriented, high-power distance, and high-context cultures. By contrast, US banks are more likely to tweet messages of <sc xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">service</small> . Overall, their messages are more likely to be characterized by indirectness, expressiveness, and involvement politeness. Specifically, they tend to send <sc xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">service</small> tweets indirectly and in the involvement politeness style. As such, US banks may best be described as financial product sellers and considerate service providers, a combination that suggests customer-oriented values in individualism and high uncertainty avoidance cultures. In addition, both Chinese and US banks claim to be good corporate citizens. This comparative analysis sheds light on identity construction and provides instructive frameworks for cross-cultural communication on social media platforms.

Micro‐blog sentiment classification using Doc2vec + SVM model with data purification

As a Chinese version of twitter, micro-blog has been popular for many years. On this platform, a lot of comments are generated explosively every day. These comments contain the public's opinions on various topics, which have wide applications in both academic and industrial fields. In recent years, deep learning and some classification algorithms have been applied to sentiment analysis, and good results have been achieved. However, micro-blog sentiment classification is a challenging task, because micro-blog messages are short and noisy, and contain massive user-invented acronyms and informal words. Unfortunately, most researchers pay more attention to analyse the data after deep learning, but only simply remove the noisy data before using algorithm, so the result of sentiment analysis has reached a bottleneck. Here, the authors first purify the data using varied methods before deep learning, then, the Support Vector Machine (SVM) classification algorithm is applied to sentiment classification of micro-blog using many types of features. Through comparing with the method of simply pre-processing data, the results show that their approach can improve the performance of micro-blog sentiment classification effectively and efficiently.

Sherloc: a knowledge-driven algorithm for geolocating microblog messages at sub-city level

ABSTRACT Many solutions for coarse geolocating of users at the time they post a message exist. However, for many important applications, like traffic monitoring and event detection, finer geolocation at the level of city neighborhoods, i.e., at a sub-city level, is needed. Data-driven approaches often do not guarantee good accuracy and efficiency due to the higher number of sub-city level positions to be estimated and the low availability of balanced and large training sets. We claim that external information sources overcome limitations of data-driven approaches in achieving good accuracy for sub-city level geolocation and we present a knowledge-driven approach achieving good results once the reference area of a message is known. Our algorithm, called Sherloc, exploits toponyms in the message, extracts their semantic from a geographic gazetteer, and embeds them into a metric space that captures the semantic distance among them. We identify the semantically closest toponyms to a message and then cluster them with respect to their spatial locations. Sherloc requires no prior training, it can infer the location at sub-city level with high accuracy, and it is not limited to geolocating on a fixed spatial grid.

Analyzing Microblogging Activity and Stock Market Behavior through Artificial Neural Networks

This paper attempts to analyze the relationship between social network activity (message sentiment) and stock market (trading volume and risk premium). We used Artificial Neural Networks to analyze 87,511 stock-related microblogging messages related to S&amp;P500 Index posted between October 2009 and October 2014. The results obtained suggest that there is a direct relationship between trading volume and negative sentiment, and between risk premium and negative sentiment. The paper concludes with several directions for future research.

What You Say and How You Say it: Joint Modeling of Topics and Discourse in Microblog Conversations

This paper presents an unsupervised framework for jointly modeling topic content and discourse behavior in microblog conversations. Concretely, we propose a neural model to discover word clusters indicating what a conversation concerns (i.e., topics) and those reflecting how participants voice their opinions (i.e., discourse).1Extensive experiments show that our model can yield both coherent topics and meaningful discourse behavior. Further study shows that our topic and discourse representations can benefit the classification of microblog messages, especially when they are jointly trained with the classifier.Our data sets and code are available at: http://github.com/zengjichuan/Topic_Disc .

Sentiment Analysis on Microblogging with K-Means Clustering and Artificial Bee Colony

Microblogging is a type of blog used by people to express their opinions, attitudes, and feelings toward entities with a short message and this message is easily shared through the network of connected people. Knowing their sentiments would be beneficial for decision-making, planning, visualization, and so on. Grouping similar microblogging messages can convey some meaningful sentiments toward an entity. This task can be accomplished by using a simple and fast clustering algorithm, [Formula: see text]-means. As the microblogging messages are short and noisy they cause high sparseness and high-dimensional dataset. To overcome this problem, term frequency–inverse document frequency (tf–idf) technique is employed for selecting the relevant features, and singular value decomposition (SVD) technique is employed for reducing the high-dimensional dataset while still retaining the most relevant features. These two techniques adjust dataset to improve the [Formula: see text]-means efficiently. Another problem comes from [Formula: see text]-means itself. [Formula: see text]-means result relies on the initial state of centroids, the random initial state of centroids usually causes convergence to a local optimum. To find a global optimum, artificial bee colony (ABC), a novel swarm intelligence algorithm, is employed to find the best initial state of centroids. Silhouette analysis technique is also used to find optimal [Formula: see text]. After clustering into [Formula: see text] groups, each group will be scored by SentiWordNet and we analyzed the sentiment polarities of each group. Our approach shows that combining various techniques (i.e., tf–idf, SVD, and ABC) can significantly improve [Formula: see text]-means result (41% from normal [Formula: see text]-means).

Geovisual analysis of VGI for understanding people's behaviour in relation to multifaceted context

Geovisual analysis of VGI for understanding people's behaviour in relation to multifaceted context

Joint Modeling of Characters, Words, and Conversation Contexts for Microblog Keyphrase Extraction

Millions of messages are produced on microblog platforms every day, leading to the pressing need for automatic identification of key points from the massive texts. To absorb salient content from the vast bulk of microblog posts, this article focuses on the task of microblog keyphrase extraction. In previous work, most efforts treat messages as independent documents and might suffer from the data sparsity problem exhibited in short and informal microblog posts. On the contrary, we propose to enrich contexts via exploiting conversations initialized by target posts and formed by their replies, which are generally centered around relevant topics to the target posts and therefore helpful for keyphrase identification. Concretely, we present a neural keyphrase extraction framework, which has 2 modules: a conversation context encoder and a keyphrase tagger. The conversation context encoder captures indicative representation from their conversation contexts and feeds the representation into the keyphrase tagger, and the keyphrase tagger extracts salient words from target posts. The 2 modules were trained jointly to optimize the conversation context encoding and keyphrase extraction processes. In the conversation context encoder, we leverage hierarchical structures to capture the word‐level indicative representation and message‐level indicative representation hierarchically. In both of the modules, we apply character‐level representations, which enables the model to explore morphological features and deal with the out‐of‐vocabulary problem caused by the informal language style of microblog messages. Extensive comparison results on real‐life data sets indicate that our model outperforms state‐of‐the‐art models from previous studies.

Emotional, especially negative microblogs are more popular on the web: evidence from an fMRI study.

Microblogs are one of the main social networking channels by which information is spread. Millions of users repost information from microblogs and share embedded emotion at the same time. The present study employed a mimicked interface of microblog (sina Weibo) and recruited university students to investigate users' propensity to repost microblogs of positive, negative or neutral valence, and studied the neural correlates with reposting microblogs of different emotional valence. Ninety pieces of microblog messages, consisting of 30 positive, 30 negative and 30 neutral were read by 28 participants (14 males and 14 females). Their propensity to repost, valence-related neural activity, and reposting-related neural activity were recorded when they read messages and decided whether to repost them. We found reposting behavior was moderated by emotion. Participants preferred to reposting emotional microblogs relative to neutral microblogs, corresponding to amplified activity in dorsolateral prefrontal cortex (DLPFC), insula, precuneus and tempoparietal junction (TPJ), the key nodes of cognitive control, emotion, self-relevance processing and mentalizing respectively. Moreover, negative microblogs were more reposted than positive ones, corresponding to amplified activity in postcentral gyrus, superior frontal gyrus (SFG) and TPJ. Reposting negative microblogs induced increased activity in bilateral TPJ, indicating that TPJ plays a key role on decisions to propagate negative information. These findings reveal an important behavioral pattern in which negative information prevails in the transmission through microblogs, as well as the neural correlates with this process. It also provides empirical evidence of how reposting works in microblogs and how the brain is involved in social propagation of information.

A Joint Model of Conversational Discourse and Latent Topics on Microblogs

Conventional topic models are ineffective for topic extraction from microblog messages, because the data sparseness exhibited in short messages lacking structure and contexts results in poor message-level word co-occurrence patterns. To address this issue, we organize microblog messages as conversation trees based on their reposting and replying relations, and propose an unsupervised model that jointly learns word distributions to represent: (1) different roles of conversational discourse, and (2) various latent topics in reflecting content information. By explicitly distinguishing the probabilities of messages with varying discourse roles in containing topical words, our model is able to discover clusters of discourse words that are indicative of topical content. In an automatic evaluation on large-scale microblog corpora, our joint model yields topics with better coherence scores than competitive topic models from previous studies. Qualitative analysis on model outputs indicates that our model induces meaningful representations for both discourse and topics. We further present an empirical study on microblog summarization based on the outputs of our joint model. The results show that the jointly modeled discourse and topic representations can effectively indicate summary-worthy content in microblog conversations.

Using frame-based resources for sentiment analysis within the financial domain

User-generated data in blogs and social networks have recently become a valuable resource for sentiment analysis in the financial domain, since they have been shown to be extremely significant to marketing research companies and public opinion organizations. In order to identify bullish and bearish sentiments associated with companies and stocks, we propose a fine-grained approach that returns a continuous score in the $$[-\,1,+\,1]$$ range. Our supervised approach leverages a frame-based ontological resource which produces feature sets such as lexical features, semantic features and their combination. One of the outcome of our analysis suggests that the frame-based ontological resource we have used might be successfully applied for sentiment analysis within the financial domain achieving better results than traditional sentiment analysis methods that do not embody semantics. We also show the higher performance of a fine-grained approach based solely on the evaluation of specific substrings of the message, rather than on features extracted from the whole text of a financial microblog message through the frame-based ontological resource. We have also compared our system with semi-supervised and unsupervised approaches and results indicate that our approach outperforms the others. Last but not the least, our approach is general and can be applied on top of any existing supervised method of polarity detection.