Search

Controlling rumors is related to the interplay between the rumor and rumor refutation, which compete to attract the unaware. And rumor refutation preys on the rumors, and try to eliminate rumors. Therefore, there is both a predatory and competitive relationship between the rumors and the rumor refutations. This paper presents a model based on biomathematics theory to describe the interplay between rumors and rumor refutations. The theoretical analysis of the differential equations elucidated three dynamic cases: rumor extinction; rumor refutation extinction; and rumor and rumor refutation coexistence. The subsequent analysis of the equilibrium stability in the three cases revealed both equilibrium stability and model instability. The haze rumor and the official haze refutation data were then crawled from Sina microblog, the rumor propagation process analyzed, an integral method employed for the model parameter estimation, and the proposed model used to estimate and forecast the haze rumor and rumor-refutation evolutions. The findings suggested that rumors and rumor refutations coexist, which was consistent with the theoretical analysis of coexistence. Based on this case and to more deeply analyze the effect of an authority’s credibility and the public’s cognition, the proposed model was used to suggest several scenarios, and policy suggestions given to assist authorities better manage rumors in emergency events.

Social networks are becoming increasingly collaborative and interdependent for achieving a broader market and profits. However, the coupled social networks which caused by the collaboration and interdependence bring more significant security risks. Rumors will spread more effectively and more rampant in coupled social networks than in separate social networks. Rumors spread in a single social network have attracted considerable attention. In coupled networks, whereas, the research remains to be explored. In coupled networks, in view of the fact that the public will face more rumors, this paper proposes a SIHR rumor spreading model with the “hesitation” psychological state for studying the dynamic course of rumors. The mean-field equation is given for describing and analyzing the diffusion course. The various propagation performances of SIHR model are explained by theoretical and experimental analysis. At last, targeted security scheme is discussed.

With the rapid development of social networks, an increasing number of factors will affect the spread of rumors. In this paper, we simultaneously consider individual activity differences and rumor-refuting nodes, and we propose a new multifactor model that can be used to better analyze the process of rumor spreading in practice. The equilibrium point and the basic regeneration number R0 of the model are calculated via the differential dynamics method and the next generation matrix method. Moreover, the local asymptotic stability of the equilibrium of the model is determined based on the Routh-Hurwitz criteria. The global stability is proved via the geometric method which is based on the use of a higher-order generalization of Bendixson’s criterion. Finally, the theoretical results are verified via numerical simulation in a homogeneous network, and the effects of individual activity and the refutation mechanism on the rumor spreading process are analyzed. We find that the proposed mechanism of rumor refutation can effectively reduce the density of spreaders and the rumor duration.

Rumor stance classification is a task for identifying different stances about specific social media posts, and it is considered as an important step to prevent rumors from spreading. However, most previous studies for the rumor stance classification, which focus on content features of posts, ignore the textual information granularity and reading (or writing) habits of the public. In social networks, people may have different stances towards unverified posts with different reading habits. Based on this observation, we propose a Stochastic Attention Convolutional Neural Net (SACNN) under different textual information granularity to catch different habits of the public for the rumor stance classification task. Specifically, being treated as view-windows of readers, convolutional kernels in a SACNN contain trainable convolutional kernels and stochastic untrainable convolutional kernels, and different sizes of the stochastic convolutional kernels are used to simulate casual online reading habits and to extract course-grained features such as phrase features. After the convolutional layer, fine-grained features such as keywords can be extracted by a pooling layer. The experiments show that the average accuracy and F1 score of our proposed model are respectively 0.26% and 0.46% higher than the state-of-the-art results on the rumor stance classification task on PHEME dataset.

There have been many efforts to detect rumors using various machine learning (ML) models, but there is still a lack of understanding of their performance against different rumor topics and available features, resulting in a significant performance degrade against completely new and unseen (unknown) rumors. To address this issue, we investigate the relationship between ML models, features, and rumor topics to select the best rumor detection model under specific conditions using 13 different ML models. Our experiment results demonstrate that there is no clear winner among the ML models in all different rumor topics with respect to the detection performance. To overcome this problem, a possible way is to use an ensemble of ML models. Although previous work presents an improved detection of rumors using ensemble solutions (ES), their evaluation did not consider detecting unknown rumors. Further, they did not present nor evaluate the configuration of the ES to ensure that it indeed performs better than using a single ML model. Based on these observations, we propose to evaluate the use of an ES by examining their unknown rumor detection performance compared with single ML models but as well as different configurations of the ESes. Our experimental results using real-world datasets found that an ES of Random Forest, XGBoost and Multilayer perceptron overall produced the best F1 score of 0.79 for detecting unknown rumors, a significant improvement compared with a single best ML model which only achieved a 0.58 F1 score. We also showed that not all ESes are the same, with significantly degraded detection and large variations in performance when different ML models are used to construct the ES. Hence, it is infeasible to rely on any single ML model-based rumor detector. Finally, our solution also performed better than other recent detectors, such as eventAI and NileTMRG that performed similar to using a single ML model - making it a much more attractive solution to detect unknown rumors in practice.

Exploring the subjective and objective factors that influence the spreading behavior of rumors can realize rumor governance. Taking into account the two types of rumors spreaders currently ubiquitous-content-driven communicators and interest-driven broadcasters, as well as the two factors affecting the content of the rumors and the interests of the spreaders, this article improves the traditional BASS model and establishes the relationship function between the subjective and objective influencing factors. Through experimental simulation, the influence of the importance and ambiguity of the rumors, the psychological benefits of the spreader and the expected attention on the number of rumors spreading is analyzed. This article analyzes the subjective and objective influencing factors of rumor spread, reveals the internal laws, and provides a theoretical basis for the proposal of rumor control strategies.

Online rumor supervision is one of the core issues in rumor governance. Based on the stakeholder theory, this study identified three main stakeholders in rumor supervision, and considered the collusion between the insider and the media, and constructed a tripartite evolutionary game model between the insider, the media and the government regulatory agency. The evolutionary stability of each participant's strategy selection is analyzed by using the replication dynamic equation.

With the rapid development of social media such as micro-blog, the rapidly spreading rumors have become an increasingly prominent problem. How to quickly and effectively identify and stop rumors has become a hot issue in the field of information dissemination. This paper introduces the concept of rumor identification and detection and common machine learning algorithms, analyzes the main methods of micro-blog rumor identification and the application process of rumor identification based on machine learning, and introduces the application example of micro-blog rumor detection based on machine learning for research prospect.

Fake News detection is much needed in today's world as it has a large impact in our social as well as personal life but involves some challenges due to the limited resources like datasets, research papers. Nowadays lot of information is being shared over social media and people cannot able to differentiate between which information is fake and which is real. People immediately start sharing the news without confirming its authenticity. This further results in spreading of it. Fake news and rumor are the most popular forms of false and unauthenticated information and should be detected as soon as possible for avoiding their dramatic consequences. In previous research, many fake news detection methods were proposed. In this project the final report is generated by predictions of text classification using SVM, Logistic Regression, Gaussian Naive Bayes and tampered image classification by Error Level Analysis using CNN.

Today, incentive policies have been proposed to contribute to small consumers in the market to improve network security. In this paper, the simultaneous competition of small consumers in the form of independent firms and power plants in the new market structure is modeled and the impact of competition of these firms on the behavior of generation firms and network costs is investigated. Given the multiplicity of consumer presence in this market and the rumor that this may slow down the market settlement process, the effect of imposing restrictions on a strategy change for all participants has been investigated.

Social media has became a critical manner for people to acquire information in daily life. Despite the great convenience, fake news can be widely spread through social networks, causing various adverse effects on people's lives. Detecting these fake news or misinformations has proved to be a critical task and draws attentions from both governments and individuals. Recently, many methods have been proposed to solve this problem, but most of them rely on the body content of the news, ignoring the social context information such as the comments. We argue that the comments of a specific news contain common judgements of the whole society and could be extremely useful for detecting fake news. In this paper, we propose a new method TRANSFAKE which jointly models the body content and comments of news systemically, and detects fake news with multi-task learning framework. TRANSFAKE model is a Transformer-based model. It takes different modalities as input and employs multiple tasks, i.e. rumor score prediction and event classification, as intermediate tasks for extracting useful hidden relationships across various modalities. These intermediate tasks promote each other and encourage TRANSFAKE making the right decision. Extensive experiments on two standard real-life datasets demonstrate that TRANSFAKE outperforms state-of-the-art methods. It improves the detection accuracy by margins as large as ~12.6% and F1 scores as large as ~15%.

Rumor propagation is becoming easier and leads to severe consequences for society in several minutes or hours due to the rapid development of social networks. Thus, detecting rumors in early time is necessary and urgent for the community. In recent studies, machine learning approaches are widely applied in detecting rumors based on various features extracted from content, user characteristics, and propagation structure. Some studies have shown that features extracted from users are more valuable for rumor early detection. Whereas existing studies utterly utilize various user features, and all of them are deemed as equally, which ignore the inter-feature and temporal difference. Therefore, in this paper, we analyze the effectiveness of six frequently used user features with several representative deep learning models to learn more about such difference. And we propose a novel annealing attention model based on the analysis. The proposed model learns feature-attention and temporal-attention with multi-layer perceptron and parameterized annealing function to capture the difference and enhance the original user features in rumor early detection. Experimental results on the real-world dataset demonstrate that the proposed model detects rumors with an accuracy of 93.6% on Weibo in 15 minutes, which outperforms the state-of-art methods.

In the field of rumor detection, the existing Transformer-based methods ignore the location information and fail to effectively use the potential information of the text. Therefore, we propose a social media rumor detection method based on improved Transformer that improves the standard Transformer through two novel techniques. First, learnable relative positional encoding is used to endow the Transformer with the ability of direction- and distance-awareness. Second, absolute positional encoding is used, through which each word with different absolute positions is mapped to its corresponding representation space. The experimental results show that, compared with the current best benchmark method, the accuracy of this method on the three data sets of Twitter15, Twitter16 and Weibo has increased by 0.9%, 0.6%, and 1.4%, respectively. The improved Transformer is effective and can significantly improve the effect of social media rumor detection.

The evolution of the World Wide Web and the rapid enhancement of social sites like Facebook, Twitter changed the way of communication. With these social media sites, people are creating and disseminating more information, meanwhile large rumor and fake news also increased. Automatic text classification as information and disinformation is a challenging task. We investigate the rumor identification problem by considering the contextual information. The proposed work is the hybridization of CNN and BILSTM with Glove embedding to classify the tweets into rumor and non-rumor. All the experiments are performed on publicly available dataset collected from Kaggle, that is the world largest community. Experimental result show that the proposed model outperformed when compared with baseline model. The proposed model provide 90.93% accuracy.

A rapid use of the social network for communication prompts the broad and quicker dispersion of information than normal news channels. The validation of information is very challenging due to massive data on a social network. Unverified news can be a talk or phony news that harms an individual, association and severe effect on humankind. Hence, battle the talk dissemination to limit the unfriendly impacts on society. Regardless of incredible endeavors to manage this issue, analysts chiefly focused on transient dynamics of posts alongside different highlights from a user, content-based, network-based features and exhibited a moderate exactness. The time series highlights are related to an occasion that smothers the other quality highlights identified with each post. There is a degree of development in precision, so this paper centers around the significant role of various features in rumor detection in social networks. This involves the highlights from the user-based, syntactic-based, lexical and temporal features. The obtained results compared to a benchmarked real-world dataset of a Twitter network.

There have recently been many cases of unverified or misleading information circulating quickly over bogus web networks and news portals. This false news creates big damage to society and misleads people. For Example, in 2019 there was a rumor that the Padma Bridge of Bangladesh needed 100,000 human heads for sacrifice. This rumor turns into a deadly position and this misleading information takes the lives of innocent people. There is a lot of work in English but a few works in Bangla. In this study, we are going to identify the fake news from the unconsidered news source to provide the newsreader with natural news or real news. The paper is based on the combination of convolutional neural network (CNN) and long short term memory (LSTM) where CNN is used for deep feature extraction and LSTM is used for detection using the extracted feature. The first thing we did to deploy this piece of work was data collection. We compiled a data set from websites and attempted to deploy it using the methodology of deep learning which contains about 50k of news. With the proposed model of Multichannel combined CNN-LSTM architecture, our model gained an accuracy of 75.05% which is a good sign for detecting fake news in Bangla.

In this paper, we propose a multimodal fusion network (termed as MFN) to integrate the text and image data from social media for rumor detection. Given the multimodal features, MFN exploits self-attentive fusion (SAF) mechanism to conduct feature-level fusion by assigning corresponding weights to the complementary modalities. In particular, the textual features are combined with the fused features in a skip-connection manner, as textual features tend to be more distinguishable compared with visual features. Furthermore, MFN introduces latent topic memory (LTM) to store the semantic information about rumor and non-rumor events, benefiting to the identification of the upcoming posts. Extensive experiments conducted on two public datasets show that the proposed MFN outperforms the state-of-the-art approaches.

Online social networks have become extremely important in daily life and can be used to influence lives in dramatic ways. Two issues are the veracity and provenance of posted information, including rumors. There is a need for methods for tracing rumors (or any piece of information) to their most likely source in such networks. We consider the detection problem of single rumor source based on observed snapshots based on the susceptible-exposed-infected-recovered (SEIR) model. According to the SEIR model, all nodes in the network are formulated into four possible states: susceptible (S), exposed (E), infected (I), and recovered (R). Given an observed snapshot in the network, from which we can know the relevant graph topology and all infected nodes, but where nodes in susceptible, exposed, or recovered status cannot be distinguished, the purpose of our research is to identify the rumor source based on the observed snapshot and graph topology. We propose the concept of the optimal infection process and derive an estimator for the rumor source based on this optimal infection process. Subsequently, we prove that this estimator matches the rumor source with a high probability. The effectiveness of the proposed scheme is validated using experiments based on regular tree networks with different degrees. We further evaluate the performance of our scheme on two well-known synthetic complex networks and four real-world networks; the results suggest that our proposed scheme outperforms the traditional rumor centrality heuristics. The performance analysis on computational complexity demonstrates that our scheme has advantages in efficiency compared with other rumor centrality heuristics used in rumor detection methods.

In this paper, incubation mechanism and activity of nodes are introduced to study the dynamics mechanism of rumor spreading in complex network. It is assumed that when the susceptible interacts with the rumor believers, the susceptible need time to think about the rumor before they believe it. The susceptible individual will first go through a incubation period after being infected before becoming believer or stifler. The nodes have active and the inactive states, it can convert each other with a certain probability, the active nodes can interact with others which directs linking, and the inactive nodes just response the active neighbors. The dynamic rumor spreading model was built by mean-field equations in heterogeneous network, and then the numerical simulations were proposed which reveals that the joint efforts of activity rate and spreading rate have negative effects on final rumor spreading size, and the simulations results also confirm the theoretical analysis in the modified model.

The advent of social media and technologies augmenting social communication has dramatically amplified the role of rumor spreading in shaping society, via means of misinformation and fact distortion. Existing research commonly utilize contagion mechanisms, statistical mechanics frameworks, or complex-network opinion dynamics models. In this paper, we incorporate information distortion and polarization effects into an opinion dynamics model based on information entropy, modeling imprecision in human memory and communication, and the consequent progressive drift of information toward subjective extremes. Simulation results predict a wide variety of possible system behavior, heavily dependent on the relative trust placed on individuals of differing social connectivity. Mass-polarization toward a positive or negative consensus occurs when a synergistic mechanism between preferential trust and polarization tendencies is sustained; a division of the population into segregated groups of different polarity is also possible under certain conditions. These results may aid in the analysis and prediction of opinion polarization phenomena on social platforms, and the presented agent-based modeling approach may aid in the simulation of complex-network information systems.

Interactions among different parties within social networks are greatly dependent on trust. Therefore, trust analysis is significant for solving social network related problems such as privacy protect, and rumor tracking and containment. This paper makes advancements in the trust analysis by proposing a reliability model-based algorithm for assessing the trust level of any two parties within a social network. Particularly, a multi-level trust model with the probability distribution is proposed and a multivalued decision diagrams (MDD)-based method is suggested for assessing the trust level of two parties that may be connected through multiple indirect or direct links. These connection paths may be correlated due to sharing a common party or link. Further, the MDD-based method is extended for performing a trust sensitivity analysis with the aim to pinpoint which direct link contributes the most to the trust relationship between two non-directly connected parties within the social network. Dynamics in trust are also investigated. Examples are provided to illustrate the proposed probabilistic MDD-based method for trust and sensitivity analyses. Performance of the proposed method is evaluated through experiments and comparisons with existing trust evaluation methods.

The convenience of social media in communication and information dissemination has made it an ideal place for spreading rumor events, which raises a higher requirement for automatic debunking of rumor events. Meanwhile, the traditional rumor classification approaches relying on manual labeled features have to face a daunting number of manual efforts. In general, when facing a dubious claim, people can authenticate and verify the realness of an event with the contents of continuous posts, such as source credibility, public sentiments, propagation structures, and so on. In this paper, we pay more attention to the emotional expressions of posts host, especially the fine-grained sentiments, which are effective for rumor events detection. Thus, this paper presents a novel two-layer GRU model for rumor events detection based on a Sentiment Dictionary (SD) and a dynamic time series (DTS) algorithm, named as SD-DTS-GRU. The model learns continuous representations of microblog events in a better manner by making use of the SD to identify fine-grained human emotional expressions of each event and retaining the time distribution of social events by the DTS algorithm. The experimental results on Sina Weibo datasets show that our model achieves a high accuracy of 95.2% and demonstrate that our proposed SD-DTS-GRU model outperforms latest explorations on rumor events detection.

We discuss the problem of locating the rumored origin in the social networks. Earlier works focused on network structure and sensor nodes; we concentrate on the timestamp of infected nodes. We recommend a two-stage algorithm to find the root node in a network using neighbor nodes' timestamp. In the first stage, discover the neighbor nodes with the minimal timestamp. Second, find the origin node using the minimal timestamp of the friend nodes. Simulations illustrate that our proposed two-stage algorithm works well and determines the root node within few hops. We evaluate our algorithm on small- scale networks, large-scale networks, and artificial networks. Our numerical results also show that our recommended two-stage algorithm operates well to discover the root node accurately, with a higher success rate and fewer distance error hops.

The widespread use of social media has had a terrible impact on our society due to the spread of fake news. In particular, the author has seen the idea of the quality of fake news before its origin. Like the internet, publishers used false and misleading information to further their interest. People often get involved in social media as social media provides low cost, quick access, and fast spread of news. It has been seen for many years that fake news harms persons as well as society. So, the challenge of fake news detection arrived. Inappropriate news took place to attract people so the sender can start putting the rumor of news. This had led to a negative impact on people about social media for their news. This led to inconvenience for offline news as well because when people will too much depend on online platforms that will reduce the offline users. This survey deals with a review of existing machine learning algorithms Naive Bayes, Convolutional Neural Network, LSTM, Neural Network, Support Vector Machine proposed for detecting and reducing fake news from different social media platforms like Facebook, whatsapp, twitter, etc. This review provides a comprehensive detail including data mining perspective, evaluation metrics, and representative datasheets. Further, a comparison of the state-of-the-art is presented and untackled challenges in detecting fake news are highlighted. Research in the field of detecting fake news has been hampered to a great extent by the lack of quantity and quality of existing datasets. Therefore, this review compares the existing approaches to build models and with further improvements to be expected by using the combination of different machine learning techniques.

During the period of epidemic prevention and control, the spread of negative and false information of public opinion may affect the prevention and control work, and even affect the image of individuals, enterprises, units and the country, so the prevention and control of public opinion is of great significance. Using big data analysis methods, we can detect public opinion under the background of the epidemic more quickly and accurately, so as to guide the bad public opinion in time. This paper summarizes the specific steps and analysis methods of network public opinion monitoring for big data, gives some hot events of epidemic situation, and introduces automatic rumor recognition. The effect of automatic rumor recognition varies with the data set, and the accuracy rate of epidemic rumor recognition based on BERT model has reached 86.7%. This paper also analyzes the psychological characteristics of different groups of Internet users during the epidemic period, and puts forward overall suggestions for the public opinion control of emergencies in the epidemic.

Due to the gradual increase of the amount of information carried by the Internet, the content of the information is uneven, and various kinds of network rumors have followed one after another, with various forms, diverse communication channels and concentrated focus problems, which have brought varying degrees of harm to all sectors of society. With respect to the risks caused by Internet rumors and serious social problems reflected in them, it is imperative to propose the targeted countermeasures. [Objectives] The purpose is to comprehensively summarize the existing problems reflected by the data in the process of Internet rumor governance and put forward corresponding solutions to deal with the complicated problems in the governance orientation of Internet rumor risks. [Methods] A field survey was conducted in the network information department of the local municipal party committee and the network supervision department of the public security bureau. Based on the data obtained, the types and the criterion of network rumors were summarized to find effective countermeasures for network rumors. [Results] According to the research results, three dilemmas in the process of controlling Internet rumors at the present stage are proposed, namely, how to trace the source, how to correct and how to improve efficiency. Based on the perspective of common governance, various paths are proposed, such as supporting the government to build a multi-level network public opinion guidance mechanism, supporting new system optimization to trace the source and identify rumors, and strengthening network self-discipline to lower the threshold of efficiency,[4] so as to cope with some weak links under the plight of rumor governance, and provide reliable countermeasures for government governance.

Amid a world of the COVID-19 pandemic, people often have turned to social media as a means of communicating and receiving information. Within existing social networks one can discover a wide range of messages that vary in sentiment, measure of helpfulness, and/or intended harm. In this paper, we measure the utility of Tweets based on their ability to propagate throughout the Twittersphere as well as the salient features that contribute to their successful propagation. In this work, we consider five types of Tweets that can be classified as: (1) Caution & Advice, (2) Doubt & Criticism, (3) Rumor & Counter-Rumors, (4) Generic Harm, and (5) Non-Situational Information. Our results suggest that, on average, Caution & Advice messages propagate fastest while messages classified as Generic Harm have the slowest propagation.

Internet rumors are confusing and destructive, and it is difficult to identify them from texts, this is why rumors are flying all over the place. The rapid identification of Internet rumors can effectively control their spread and achieve good public governance. Considering the similarity between multiple rumor texts, based on the fractal theory, a fractal dimension calculation method for Internet rumors covered by pixel points is constructed. The Internet rumor text is divided into keywords and non-keywords to form a binarization matrix, so as to build a two-dimensional image with the same resolution, in order to realize the rapid recognition of rumors. When using the Internet rumors title text as a sample for model verification, the box-counting dimension solved by different matrices constructed by a rumor is quite different because of the effect of the increased pixel points when the equal-length heading text generates different matrices, that is, the shape of the two-dimensional image has a great influence on the fractal dimension. Therefore, it is difficult to realize the recognition of Internet rumors using the box-counting dimension algorithm based on the pixel coverage method.

In this paper,the fractal characteristics of the rumor spreading are studied by the perspective of time series constructed with the influence of rumor spreading.The values of the Hurst exponent calculated through rescaled range analysis indicate that the time series of rumor spreading are fractal with long-range dependence and self-similarity.The visual method is used to convert the time series into a complex network. The scale-free and small-world nature of the network further verify that the time series of rumor spread has fractal characteristics. These conclusions will help scholars to study the spread of rumor from a new perspective and provide ideas for the governance of rumor.

Social networking, information sharing, knowledge imparting, discussions on current happenings etc. are always a part of human society. With the fast pace of life and advancement in technology; people rely more on online information, as a result of this web platforms have become a dominant place for social interactions. This has given rise to unverified and unauthenticated news that has extremely negative effects. Fake news, rumor, misinformation, disinformation, satire, hoax, clickbait, propaganda are all different flavors of the same malice of information pollution. The research community is constantly trying to figure out a viable technical solution to this problem in different ways. In this work, we designed a framework based on five independent supervised machine-learned classifiers Support Vector Machine, K-Nearest Neighbor, Logistic Regression, Naive Bayes and Random Forest for trustworthiness assessment of web information contents. The classifiers are being trained and tested on two different datasets: Fake News Detection (Jruvika/FND) and Real or Fake News that contains full news articles in the form of headline and body. Experiments and result analysis verify that the highest accuracy attained by the projected method is 96.61% on the Fake News Detection dataset using the SVM classifier. The work is also compared with other contemporary techniques.

This research investigates how Internet users comment in response to rumor corrections posted on social media. The focus is specifically on the degree to which aggressive language is used. As the test cases for investigation, the research looks into two rumor corrections on YouTube. The rumors were set in the context of the riots and protests in Jakarta following the Indonesian presidential election in 2019. A total of 1,000 comments (500 comments from each of the two cases) was admitted for content analysis. In one case, anti-correction voice was dominant, highlighting the failure of the rumor correction to refute the rumor. In the other, pro-correction voice was dominant, indicating the success of the rumor correction. Aggressive language was widely used in the latter. Implications of the findings are highlighted.

The past decade has witnessed a rapid growth in the use of online social networks, such as Twitter, by individuals as well as communities for fast dissemination of information. However, as useful as this information might be, it also prospects the rapid spread of rumors. This phenomenon of rumor proliferation has persuaded researchers to work in the field of rumor detection by using the temporal, textual and author related features. From a sociological perspective, rumors are generally seen to be targeted at some important entities such as well-known politicians, actors, places etc. Some rumors are aimed at creating social unrest by spreading information about eye-catching events like highjacking, bomb attack etc. The existing works in rumor detection do not seem to harness this peculiar characteristic of rumor.This research proposes a novel approach to rumor detection by using entity recognition on the post text along with other features that imply reliability and consistency. Our hypothesis defines four different scores namely, i) Famousness score, ii) Rareness score, iii) Reliability score and iv) Consistency score. Experimental results show that the proposed features not only increase the performance of our model, but also out-perform the baseline approaches in terms of F1 score.

With the increasing use of online social networks as a source of news and information, the propensity for a rumor to disseminate widely and quickly poses a great concern, especially in disaster situations where users do not have enough time to fact-check posts before making the informed decision to react to a post that appears to be credible. At the same time, we know that misinformation is easily detectable by a certain few, very skeptical, or very informed users. In this study, we demonstrate how blending artificial intelligence and human skills can create a new paradigm for credibility prediction. The crowdsourcing part of the detection mechanism is implemented implicitly, by simply observing the natural interaction between users encountering the messages. Specifically, we explore the spread of information on Twitter at the microscopic (user-to-user propagation) level and propose a model that predicts if a message is True or False by observing the latent attributes of the message, along with those of the users interacting with it, and their reactions to the message. We demonstrate the application of this model to the detection of misinformation and rank the relevant message and user features that are most critical in influencing the spread of rumor over the network. Our experiments using real-world data show that the proposed model achieves over 90% accuracy in predicting the credibility of posts on Twitter, a significant boost over state-of-the-art models.

Nowadays, the information about COVID-19 is spreading widely on social networks while rumors about it are also propagating quickly. When users receive two opposing opinions, they are more likely to accept the information with greater impact. Therefore, there is an urgent need to clarify rumors. Given the cost C total , the rumor clarifying problem aims to find a seed set with at most k users such that the total number of the users influenced by positive information can be maximized. In this paper, we propose a Cost-Randomized Greedy algorithm called CRG to solve the rumor clarifying problem. It runs in $O((n-r)mC_{total})$, where $n, m$ and r are the number of nodes, edges and the initial rumor nodes, respectively. Extensive contrast experiments on online networks show that our algorithm is much more efficient than the state-of-the art methods.

Using a new opinion dynamics using the polarity analysis of measured data of Twitter about "toilet paper" and "COVID-19", we analyzed the chain of fear that people feel for the rumor of lack of toilet paper and COVID-19 itself. We use opinion dynamics for the analysis of the fears, and we obtain the opinion distribution of the fear calculated using the opinion dynamics and the observed distribution. By comparing the two distribution, we found that the opinion distribution for the fear is consisted with the mixture of trust and distrust.

Most of the consumers will search for the reviews of products or service to help them make a decision to shopping or not, so they consider the authenticity of comments is very important. Moreover, company will collect positive and negative comments to realize their position within the market, and sentiment analysis is regarded as a beneficial tool. Most of the researches only focused on Identifying authenticity or sentiment analysis, our study attempted to solve both of two issues. We constructed an integrated classifier based on text mining approaches for Identifying authenticity and sentiment analysis, the data which this study used was collected by Amazon Mass Intelligence Platform and Trip Advisor hotel review website. There are two experiments tested in this study which are mixed judgment and judge by step. The mixed judgment was analyzing real vs. fake and positive vs. negative reviews into a matrix. Additionally, judge by step was analyzing the reality of reviews then conducting sentiment analysis. With the experimental results, the system will figure out the keywords through two theories--sentiment analysis and rumor/lie theory, based on sentiment analysis the comments could divide into positive and negative. Furthermore, if the user would like to know the comments are fake or not, then based on the rumor/lie theory. During the process, the experiments were not only use a single classifier in order to pick the best model but use the method of integrating to enhance the accuracy of classifier. The results of this study illustrated the comparison between judge by step and mixed judgment. The accuracy which in step method (77.29%) is better than mixed judgment (68.13%). This study found out the step methods is more complicate than mixed judgement, but the accuracy and F- measure are the better.

The new coronavirus epidemic (COVID-19) has received widespread attention, causing the health crisis across the world. Massive information about the COVID-19 has emerged on social networks. However, not all information disseminated on social networks is true and reliable. In response to the COVID-19 pandemic, only real information is valuable to the authorities and the public. Therefore, it is an essential task to detect rumors of the COVID-19 on social networks. In this paper, we attempt to solve this problem by using an approach of machine learning on the platform of Weibo. First, we extract text characteristics, user-related features, interaction-based features, and emotion-based features from the spread messages of the COVID-19. Second, by combining these four types of features, we design an intelligent rumor detection model with the technique of ensemble learning. Finally, we conduct extensive experiments on the collected data from Weibo. Experimental results indicate that our model can significantly improve the accuracy of rumor detection, with an accuracy rate of 91% and an AUC value of 0.96.