Sina Microblog Research Articles

A Study on the Evolution of Public Sentiment and Emotional Communication Mechanisms in Women's Issues on microblog: Case Study of "Quan XiXi' s Dialogue with Chizuko Ueno"

The research utilizes Sina microblog as its platform, analyzing textual data associated with trending topics featuring "Peking University Dormitory and Ueno Chat", to explore patterns of sentiment evolution and emotional expression. Findings reveal that public sentiment follows a four-phase trajectory: inception, peak, fluctuation, and decline. Negative sentiments dominate among users discussing the topic. This analysis elucidates microblog's role in disseminating women's issues, identifies key characteristics and challenges of such discussions on the platform, delineates major disagreements and focal points in user discourse, and endeavors to uncover underlying societal symptoms. The insights contribute valuable data to studies on the forms and impacts of women's issue dissemination on microblog.

Use of meishi by Chinese females in romantic conversation

AbstractThis study explores the major types and main interpersonal functions of meishi (没事, literally ‘I'm fine’) by Chinese females in romantic conversation through analyzing collected posts from Sina Microblog. Results show that meishi by Chinese females in the context of romantic relationships primarily manifests the attributes of “expressive” and “assertive” (“insincere assertive” in particular), with specific functions to express comfort (expressive), to implicitly express negative feelings (expressive), and to avoid self-disclosure of negative emotion (insincere assertive). We hold that Chinese women's use of meishi is not only a realization of gendered discourse but also has a practical function as it detects the sincerity and attentiveness of their male counterparts.

An improved ResNet method for urban flooding water depth estimation from social media images

Social media images featuring high timeliness, large data volume, and low cost, is utilized for flooding depth estimation. Common flood depth estimation methods are to segment each reference instance, resulting in local water depths and contradictory results. This study added Coordinate Attention into Residual Neural Network (ResNet), and proposed Coordinate Attention-Residual Neural Network (CA-ResNet) to acquire global submerged water depths. The 2021 Henan rainstorm in China was used as case study and a flooding dataset from Chinese Sina Microblog was constructed, with total 5676 images and 4189 flooded. All the flooded depths class 0–9 reflected from the images were annotated manually. The proposed CA-ResNet method was compared with VGGNet, ResNet, GoogleNet, and EfficientNet, both accuracy and F1 score improved by 1% − 3%. The average accuracy of level 0–––8 within the error range of CA-ResNet reached 83.14%. It is proved that CA-ResNet could achieve flood depth estimation, facilitating efficient decision support for flood responding.

Emotional contagion on social media and the simulation of intervention strategies after a disaster event: a modeling study

With the advent of climate change and the 5 G era, online communities are increasingly becoming the main medium for information dissemination after emergencies such as natural disasters. The widespread dissemination of negative online information may generate cyber violence or lead to serious adverse psychological outcomes. This study considered a natural disaster event involving avoidable deaths and child casualties as an example to identify emotional contagion and conduct simulation interventions. Data about the aftermath of the 8·13 flash flood in the Longcaogou Scenic Area, Sichuan Province, China, were derived from the Chinese Sina microblog. We analyzed key parameters and modeled them in a dynamic model. We further evaluated the effects of implementing intervention measures (such as transmission path interruption and changing the number of different emotions) on emotional spread. The overall sentiment of posters after this flood was negative, with three epidemic peaks. Negative emotions were more persistent and contagious than positive emotions. Reducing the number of negative blog posts by half could have led to a 14.97% reduction in negative comments and a 7.17% reduction in positive comments. Simultaneously, reducing the number of negative blog posts and increasing the number of positive posts would have helped reduce the relative ratio of negative to positive comments. The findings have theoretical and practical implications for developing an emotional contagion model and formulating intervention strategies to guide public opinion after an emergency that involves extensive online debate.

Automatic depression prediction via cross-modal attention-based multi-modal fusion in social networks

Depression has become a worldwide public health concern in recent years. The online social network can be served as a tool to predict latent depressed users. However, existing studies primarily use machine learning or deep learning methods on multi-modal information with simple concatenation, resulting in unsatisfactory prediction. Additionally, the lack of interpretability in extracted features is a challenge. Furthermore, the majority of these studies primarily concentrate on English social networks. To address these issues, we propose a novel unified multi-tasking training model named CrossAMF (Cross-modal Attention-based Multi-modal Fusion). CrossAMF extracts crucial multi-modal features with explainability and effectively fuses representations of different modalities. Our approach employs a sentiment filtering-based XLNet-CNN-BiGRU model and Reduce-VGGNet to extract textual and image features, respectively. We also introduce innovative user behavior metrics previously unexplored. We collect a dataset of Chinese depressed users from Sina microblog with their texts and images, including 5692 depressed users and 9121 normal users. Experimental results demonstrate the significance of each modality to our task, with the combination yielding the best performance. CrossAMF outperforms other state-of-the-art methods with 0.9097 in accuracy and 0.8915 in F1, showing the potential of application in screening out early depression automatically.

Public attention and psychological trends towards waste reduction: A large-scale data analysis based on social media

In the digital era, using social media big data to capture the true views and psychological trends shown by the public regarding waste reduction could considerably improve the formulation of targeted waste reduction policies and guide residents to participate in environmental governance from the source side. This study used big data mining technology to trawl 617,771 pieces of waste reduction text from a typical social media site (Sina Microblog). A machine learning algorithm model was used to identify the psychological and cognitive focus of the public and their preferences based on large-scale text data. The temporal and spatial differences in public attention trends, hot topic trends, and sentiment trends were also investigated. The results showed that the public attention level was related to the release of policies by government and that public attention in the southeast coastal areas was higher than that in the northwest inland areas. Moreover, waste reduction had a “working attribute” because the public paid more attention to waste reduction during working hours (i.e., 9:00–12:00 and 15:00–18:00) than during leisure hours. In terms of individual heterogeneity, males were initiators of the topic, whereas females were followers. In particular, participation by young people in waste reduction discussions was higher than for other groups. The topic analysis showed that public attention followed a cooperative construction pattern that had multiple entities, including the individual, community, and government. Overall, the public sentiment score towards waste reduction increased year by year during the study period, with positive sentiment posts accounting for over 70% of the total number of blog posts, and that the vast majority of residents had a positive attitude towards waste reduction. This study expanded current research knowledge by exploring the public attitude response to waste reduction from a social media perspective. The study will help the government to effectively intervene in public behavior tendencies toward waste reduction from the psychological perspective and provided important implications about how the government can enhance its use of social media to effectively guide public opinion and improve policies.

Sentiment analysis of Sina microblog comments related to fertility issues based on GNN-LSTM model

Sentiment analysis of Sina microblog comments related to fertility issues based on GNN-LSTM model

Coupled dynamics of information propagation and emotion influence: Emerging emotion clusters for public health emergency messages on the Chinese Sina Microblog

A major distinction between the COVID-19 pandemic and previous pandemics is the significant role of social media platforms in shaping public adherence to non-pharmaceutical interventions and vaccine acceptance. The information propagation and individual emotion communication relevant to public health messages have been promoted in the Chinese Sina Microblog, one of the most popular social platforms in China, through two major interaction mechanisms, “commenting” and “forwarding”. Here, we develop a hybrid system of macro information propagation dynamics Susceptible-Commenting-Forwarding-Immune model coupled with the micro emotion influence dynamics Hegselmann-Krause model, termed HK-SCFI coupled system, and reveal the mechanisms of individual emotion infection using the commenting and forwarding data (including time point and content) publicly available on public opinion events in the Chinese Sina Microblog. We demonstrate how our data-driven and model-informed analysis can provide quantitative insights into the co-evolution mechanism of individual emotion during information propagation, describe the emergency of emotion clusters, and predict individual emotion evolution trends. Our study shows that multiple interactive mechanisms can facilitate the propagation of public messages and shape public sentiment through online social networks. Our modeling framework can potentially be a tool for designing effective propagation strategies during public health emergencies.

Online Public Opinion Analysis Model Based on Long Short-term Memory Network and Expectation Maximization

Abstract With the rapid dissemination of online public opinion, its emotions are easily transmitted to the general public. It possesses a certain level of social mobilization capacity and can impact the stability of society. To characterize the sentiment trend of social network information and determine its influence, we have proposed a method based on long short-term memory (LSTM) networks and expectation maximization(EM). This model employs a long short-term memory network for data training, obtaining the number of positive public opinions through word-to-word matching. Based on the expectation-maximization method and Jensen’s inequality, the lower bound of the maximum likelihood function is iteratively computed, ultimately achieving convergence of this likelihood function. This convergence value is then used for sentiment analysis. Our study utilizes 10,000 valid pieces of data from the Sina microblog as experimental data. By comparing our model with the K-MEANS model and the EM model, the results indicate significant improvements in the accuracy and convergence of our model. Our research discovers that the influence of public opinion increases as the compensation value for adoption rises, and the probability of public opinion generation gradually increases with the length of user registration years, eventually slowing down.

MODELING AND ANALYZING DYNAMIC INFORMATION PROPAGATION ON SINA WEIBO IN A SEMI-DIRECTED NETWORK

In reality, the dissemination of information about COVID-19 vaccines typically involves a combination of opinion leaders and self-organizing networks, with each node being exposed to information in varying ways. However, conventional models often assume homogeneity in networks, treating all nodes as equal in terms of propagation probabilities within a fixed timeframe, thereby neglecting the inherent heterogeneity of social networks in information dissemination. To address this limitation, we propose a novel semi-directed network model, referred to as the susceptible-forwarding-immune model, which incorporates the complex structure of actual social networks and classifies nodes based on their mode of contact and the number of users they reach within a specific period. We calibrated and validated our model using real data on COVID-19 vaccine information from the Chinese Sina microblog, and our sensitivity analysis yielded insights into optimal strategies for disseminating such information.

Predicting multi-subsequent events and actors in public health emergencies: An event-based knowledge graph approach

Public health emergencies trigger series of chain reactions that have devastating impacts on society. In addition, the subsequent events and actors in public health emergencies represent comprehensive emergency scenarios. Taking this information into account, predicting subsequent events and actors could motivate governments to take necessary and effective countermeasures. Therefore, we develop a model for predicting subsequent events and potential actors, i.e., a subsequent multievent graph convolutional network (SMEGCN), by utilizing the evolutionary information of events. Specifically, we take both relational information and semantic information into consideration to achieve improved prediction performance and simultaneously predict subsequent actors in a convenient manner. Specifically, we collect data from the Sina microblog concerning the COVID-19 pandemic to form five news datasets by employing a Python-based agent to practically test the performance of our model. The agile principle is applied to identify and handle a series of subsequent events and potential actors. The results show that embedding relational information, semantic information, and context inferences into the prediction model can improve the model performance by approximately 20%. Additionally, a comparative analysis indicates that the SMEGCN model is superior to other methods in terms of predicting both subsequent events and actors. From the perspective of an example analysis, social media, especially official media accounts, stimulates interactions between governments and the public and improves the management effectiveness of governments. However, during periods of emerging public health emergencies, the most important events that should be noted diachronically are treatment, daily life guarantees, and acute and chronic disease treatments, whereas at the end of a public health emergency, the key tasks include how to revive commercial activities and improve the vaccination rate. Drawing conclusions from these discussions, the present study not only contributes to the literature on theoretically and methodologically predicting events and actors but also provides practical suggestions for managing public health emergencies.

One Case Study of Public Opinion from Micro-blog: Perspective from Negative Public Emergencies

Using the Sina Micro-blogs related to the China restaurant assault as a source, this study explores the pattern of public opinion evolution related to negative public emergencies and the reasons behind it. This paper divides the public opinion development process into four stages according to the change of micro-blog volume, namely, the gestation period, the climax period, the fluctuation period and the recession period. It also analyzes the change of public opinion focus based on the high-frequency words in different periods. The study found that surveillance videos sparked outrage during the gestation period, and the controversy about gender opposition pushed public opinion to its peak. During the fluctuating period, conscious public scrutiny could maintain the heat of the event. At the same time, the fluctuating period also showed an obvious phenomenon of emotional infection and catharsis. In addition, this paper analyzes the first police notification micro-blog of this incident, and finds the worthwhile improvements of official government micro-blog accounts in responding to such incidents.

Governance by platforms: Configuration of the relationship between Party-government and society in China

Platforms have emerged as the dominant mode of communication in China, empowering individuals to express their opinions and question the authority and credibility of the ruling party. To illustrate the impact of platforms on Party-government relations and society in China, we analyze four representative Chinese platforms: the Sina Microblog matrix of @Yinchuan for Governing Online, Shenzhen Baoan Tong APP, Shanghai Baoshan Community Link platform, and the smart party building platform in Beihai City. Drawing on these case studies, we argue that platforms not only provide channels for party control and government transformation, but also contribute to social progress by shaping the relationship between the Party-government and society, thus reshaping Chinese governance.

Identifying underlying influential factors in information diffusion process on social media platform: A hybrid approach of data mining and time series regression

As the prevailing online communications paradigm, social media platforms are considered to be the fastest medium for sharing and diffusing information. But what influences the spread of information through these platforms? The content of the post? The sentiments contained? Or the characteristics of user's behavior? To explore which factors promote the spread of information through social media, we developed a data analytics method that combines data mining with time series regression. We then applied this analytical framework to the L group Double 11 false advertising scandal, which blew up on the Sina microblog – a public hot trend that attracted the attention of millions of people. Our analysis reveals how three factors – user activity, emotional changes, and public attention – interact and the role they play in the spread of information. Among these factors, sentiment polarity and reposting are found to be the two main drivers of information diffusion. Emotional contagion accelerates the spread of information when the event first breaks (known as the accumulation period), while reposting does more to spread information once the event has gained some traction (the diffusion period). Surprisingly, the topic of public concentration in the event has a significant impact on the spread of the event in the accumulation period, but the effect shades away during the diffusion and convergence periods, i.e., the farther relations among topics are tied, the less public interest is abating on the event – a finding that is supported by cognitive load theory. However, although public attention shows little influence in the diffusion process, it does reveal how consumers shift their attention to different subtopics over time. Overall, our analysis sheds some light on how online events evolve and ‘go viral’. Notably, this study not only explores how underlying factors dynamically influence the information diffusion process, but also offers insights into how to manage information diffusion processes in practice.

Applicability Analysis and Ensemble Application of BERT with TF-IDF, TextRank, MMR, and LDA for Topic Classification Based on Flood-Related VGI

Volunteered geographic information (VGI) plays an increasingly crucial role in flash floods. However, topic classification and spatiotemporal analysis are complicated by the various expressions and lengths of social media textual data. This paper conducted applicability analysis on bidirectional encoder representation from transformers (BERT) and four traditional methods, TextRank, term frequency–inverse document frequency (TF-IDF), maximal marginal relevance (MMR), and linear discriminant analysis (LDA), and the results show that for user type, BERT performs best on the Government Affairs Microblog, whereas LDA-BERT performs best on the We Media Microblog. As for text length, TF-IDF-BERT works better for texts with a length of <70 and length >140 words, and LDA-BERT performs best with a text length of 70–140 words. For the spatiotemporal evolution pattern, the study suggests that in a Henan rainstorm, the textual topics follow the general pattern of “situation-tips-rescue”. Moreover, this paper detected the hotspot of “Metro Line 5” related to a Henan rainstorm and discovered that the topical focus of the Henan rainstorm spatially shifts from Zhengzhou, first to Xinxiang, and then to Hebi, showing a remarkable tendency from south to north, which was the same as the report issued by the authorities. We integrated multi-methods to improve the overall topic classification accuracy of Sina microblogs, facilitating the spatiotemporal analysis of flooding.

Analysis of Differences in User Groups and Post Sentiment of COVID-19 Vaccine Hesitators in Chinese Social-Media Platforms.

(1) Background: The COVID-19 epidemic is still global and no specific drug has been developed for COVID-19. Vaccination can both prevent infection and limit the spread of the epidemic. Eliminating hesitation to the COVID-19 vaccine and achieving early herd immunity is a common goal for all countries. However, efforts in this area have not been significant and there is still a long way to go to eliminate vaccine hesitancy. (2) Objective: This study aimed to uncover differences in the characteristics and sentiments of COVID-19 vaccine hesitators on Chinese social-media platforms and to achieve a classification of vaccine-hesitant groups. (3) Methods: COVID-19-vaccine-hesitation posts and user characteristics were collected on the Sina Microblog platform for posting times spanning one year, and posts were identified for hesitation types. Logistic regression was used to conduct user-group analysis. The differences in user characteristics between the various types of COVID-19 vaccine posts were analysed according to four user characteristics: gender, address type, degree of personal-information disclosure, and whether they followed health topics. Sentiment analysis was conducted using sentiment analysis tools to calculate the sentiment scores and sentiment polarity of various COVID-19 vaccine posts, and the K-W test was used to uncover the sentiment differences between various types of COVID-19-vaccine-hesitation posts. (4) Results: There are differences in the types of COVID-19-vaccine-hesitation posts posted by users with different characteristics, and different types of COVID-19-vaccine-hesitation posts differ in terms of sentiment. Differences in user attributes and user behaviors are found across the different COVID-19-vaccine-hesitation types. Ultimately, two COVID-19-vaccine-hesitant user groups were identified: Body-related and Non-bodily-related. Users who posted body-related vaccine-hesitation posts are more often female, disclose more personal information and follow health topics on social-media platforms. Users who posted non-bodily-related posts are more often male, disclose less personal information, and do not follow health topics. The average sentiment score for all COVID-19-vaccine-hesitant-type posts is less than 0.45, with negative-sentiment posts outweighing positive- and neutral-sentiment posts in each type, among which the "Individual rights" type is the most negative. (5) Conclusions: This paper complements the application of user groups in the field of vaccine hesitation, and the results of the analysis of group characteristics and post sentiment can help to provide an in-depth and comprehensive analysis of the concerns and needs of COVID-19 vaccine hesitators. This will help public-health agencies to implement more targeted strategies to eliminate vaccine hesitancy and improve their work related to the COVID-19 vaccine, with far-reaching implications for COVID-19-vaccine promotion and vaccination.

Hot-topics cross-propagation and opinion-transfer dynamics in the Chinese Sina-microblog social media: A modeling study

On social media platforms, hot topics often contain several pieces of related information that can influence internet users, generating either positive or negative opinion orientation. Some of them will choose to retain or change their original opinions after exposure to multiple related messages. To describe the opinion-transfer transient and collective behaviors in this scenario, this paper proposes an opinion-transfer susceptible-forwarding-immunized (OT-SFI) information cross-propagation model. Real multiple information in messages with opinions obtained from the Chinese Sina microblog is used for data fitting to illustrate how model parameters can be estimated and used to predict the accumulative numbers of users with a particular view. The study attempts to relate changes in group views in the network to initial opinion distribution and individuals' opinion choices at the macro level. Furthermore, the model parameters at the micro level are used to measure the probability of “retention” and “reversal” of views in events, as well as the extent to which the masses are influenced by new information views. The result illustrates that the viewpoint distribution of the initial message and the opinion selection of the new message opinion leaders play crucial roles in promoting attention to the topic and driving for a desired collective opinion.

Text recognition and analysis of network public opinion focus events of a major epidemic: a case study of "COVID-19" in Sina Microblogs.

Identifying and analyzing the public's opinion of focal events during a major epidemic can help the government grasp the vicissitudes of network public opinion in a timely manner and provide the appropriate responses. Taking the COVID-19 epidemic as an example, this study begins by using Python-selenium to capture the original text and comment data related to COVID-19 from Sina Microblog's CCTV News from Jan. 19, 2020, to Feb. 20, 2020. The study subsequently uses a manual interpretation method to classify the Weibo content and analyzes the shifting focus phenomena of network public opinion based on the moving average method. Next, the study uses an enhances TF-IDF to extract keywords from the Weibo comment and uses the keywords to construct a word co-occurrence network. The results show that during the epidemic, the network public opinion focus shifted significantly over time. With the progression of the epidemic, the focus of network public opinion diversified, and various categories stabilized. Compared to simple keyword and text classification recognition focus problems, the proposed model, which is highly accurate, identified multiple network public opinion focus problems and described the core contradictions of the different focus problems.

Causes and Countermeasures of Illegitimate Complaints from Parents in China's Education System Case Study of "A Teacher's Corporal Punishment towards Her Student Resulting in Vomiting Blood" in Weibo

In the most popular Chinese social media Sina Microblog, which is called Weibo , a netizen complained an incident that "a teacher's corporal punishment towards her student which caused vomiting blood". The incident was widely followed and spread on Weibo but then proved to be fake. Though illegitimate content was issued, from the original intention of the one involved in the claim for rights protection, the reason for this illegitimate complaining behavior is more profound. On the one hand, there exists a lack of social credibility in China's education system and the phenomenon of stigmatized teachers, which leads to an urgent voice for the supervision of educational practitioners among the masses; on the other hand, the expression of China's civil rights and its communication channels need to be further improved to promote deep development of China's civil society. Therefore, the public power of education needs to be further standardized and the multi-level system of safeguarding education rights to be optimized, so as to promote the construction of civil society and to adapt to the latest trends of social development. Meanwhile, the Internet complaining system needs to be improved which means the increase of punishment for illegitimate accusations and defamation and promotion of the Internet legal ecosystem.

The phenomenon of feminist stigmatization and the research of cyber violence against women

Due to the development of the Internet platform, feminism has gradually come into the public view, and with it comes the controversy over the topic of feminism. Taking the famous Chinese online platform Sina Micro-blog as an example, this paper analyses the current situation of feminism on the Sina Micro-blog platform and the reasons for its stigmatization, in order to further analyze the impact of the phenomenon of feminism on cyber violence against women, and to make analysis and suggestions for improving this situation and protecting women's rights.