Rumor propagation analysis during the COVID-19 outbreak

Размещено на http://www.allbest.ru/

Federal State Autonomous Educational Institution

For Higher Education

National Research University Higher School of Economics

Faculty of Communications, Media, and Design

Rumor propagation analysis during the covid-19 outbreak

Clara Christina Rondonuwu



Table of Contents

List of Figures

Abstract

Introduction

1. Literature review

2. Data and Methodologies

2.1 Research design

2.2 Case selection

2.3 Data collection

2.4 Analytical strategy

2.5 Delimitations

3. Findings

3.1 Rumor: WHO let the dogs out

3.2 Rumor: Orangutan washing hands during the pandemic

3.3 Rumor 3: Swans return to Venetian canals

3.4 Rumor 4: Drunk elephants in a tea garden

3.5 Rumor 5: Rare Malabar civet spotted in India

Conclusion

Bibliography



List of Figures

Figure 1. Extracted time series, with x-axis as days and y-axis as the number of retweets on the topics (log). Source: Author's construction 9

Figure 2. Followers compared to the number of retweets. Most users have followers between 0 to 100,000. Followers (x), number of retweets (y) 10

Figure 3. Network graphs of rumor spreading and distribution of retweets related to rumors on (a) dog ;(b) orangutan ;(c) swan ;(d) elephant; (e) civet. Each community is differentiated with color, each node sized by in-degree or the size of the audience. Source: Author's construction. 14

Figure 4. A tweet that made a pun after Baha Men hit song. Source: Author's screen capture, May 6,2020. 15

This is a long-standing rumor that circulates for three weeks. The network consists of nine groups that formed a community clusters network type (see Figure 5. for visualization). This type of network group is form around multiple news sources and popular topics, in this case human-to-animal transmission. The user in the center of this network is @DiageoLiam. Other users still retweet his tweet after 19 days. 16

Figure 6. A tweet claiming orangutan washing hands due to pandemic. Source: Author's screen capture, May 6, 2020. 18

Figure 7. A tweet claiming that swans return to Venetian canals. Source: Author's screen capture, May 6, 2020. 20

Figure 8. A tweet claiming elephants passed out after drinking corn wine. Source: Author's screen captures on May 6, 2020. 22

Figure 9. A tweet claiming that a critically endangered civet resurfaces in India during lockdown. Source: Author's screen capture on May 6, 2020. 24



Abstract

Throughout the COVID-19 pandemic, we witnessed the flourishing rumors of wild animals reclaiming urban spaces in quarantined cities. Hundreds of thousands of retweets generate in one minute, rapid amplitude builds up as people spread the information without questioning its accuracy. This dissertation uses graph visualization within a social network analysis study to explore information flow and mine propagation patterns from five popular rumors featuring animals. Ultimately, this study illustrates how information without a diffusion pattern is a set of isolated statements, whereas a diffusion pattern without information is a network without meaning. Together, information and the diffusion pattern of rumor determine the best strategy available to reduce society's vulnerability disinformation in times of high information, uncertainty, and anxiety.

Keywords: COVID-19; infodemic; rumor propagation; retweet network; Twitter.



Introduction

Social media is a great source of information. However, not all information is correct or helpful. Therefore, social media is also called the 'web of deceit' (Mintz, 2012; Wooley and Howard, 2019). As respiratory disease COVID-19 sweeps through the world, rumors become increasingly pervasive with more people retweeting, reposting, or simply liking information they find useful.

The World Health Organization expresses their concerns because false information spreads faster than truth. Many people have fallen foul in an infodemic that just 'as dangerous as the disease itself' and could exacerbate the crisis. On 15 February, the director general of the World Health Organization, Tedros Adhanom Ghebreyesus, defined:

Infodemic is an overabundace of information--some accurate and some not--that makes it hard for people to find trustworthy sources and reliable guidance when they need it (WHO, 2020).

Rumors circulating online includes conspiracy theories, speculations about coronavirus, myths, and recently viral untruths of wild animals reclaiming urban spaces in quarantined cities. There is an opportunity to study these rumors, asking what factors shape a rumor in an information crisis, particularly looking at how endogenous and exogenous pressures interact to start or stop the rumor diffusion. This study uses graph visualization within a social network analysis study to track information propagation and mine patterns from five popular rumors featuring animals, to answer the research question: How do animal rumors spread throughout Twitter during the COVID-19 information crisis?

This exploration will focus not on verifying misinformation, but rather on exploring broader factors that illustrate the underlying process of rumor propagation in Twitter social network and identifying dynamic attributes in each retweet networks that interact with information distribution at the community level during an information crisis.

While some happy rumors about drunk elephants in China, or dolphins and swans had returned to deserted Venetian canals may soften the blows of news about COVID-19 case surges, quarantine orders, and medical supply shortages, there can be harm in sharing any untruth information in Twitter social networks. Falsehood consistently dominates the truth on Twitter (Vosoughi et al., 2018). It reaches more people, penetrates deeper into the social network, and spread much faster than truth does.



1. Literature review

There is a considerable body of literature on several aspects relevant to this study. Recent work by (Nielsen et al., 2020) has attempted to explore the landscape of COVID-19 infodemic by type, source, and claims. They showed that celebrities and politicians have played a key role in spreading virus misinformation in social media. (Cinelli et al., 2020) put forward a massive data analysis of engagement on Twitter, Instagram, YouTube, Reddit, and Gab to estimates the rumors amplification during the crisis. (Pennycook, Mcphetres, et al., 2020) shows how social media fuels wave of COVID-19 misinformation as users focus on popularity, not accuracy. (Hu et al., 2020) established an infodemiological study on perception bias caused by the virus naming in the COVID-19 epidemic and infodemic. (Kucharski, 2020) shows the parallels between epidemics, recessions, and fake news.

This study sits within the study of social contagion in network communications, that takes the lessons learned from the way viruses diffuse. An important work in this area is from (Daley & Kendall, 1965), they classified individuals into ignorant, spreader, and stifler. Ignorants are those who are unaware of the rumor, spreaders are those who are aware and actively spread the rumor, and stiflers are those who have heard the rumor, but have since decided to no longer spread it. A variant of this model was given by Maki and Thompson (Maki, 1973). In this model, a rumor is spread by direct contacts of the spreaders with other in the population. Furthermore, when a spreader contacts another spreader only initiating s preader becomes a stifler.

Rumor can be defined in a number of ways. In this paper, the author frames rumor as a particular piece of information that spreads throughout a social network.

Rumor can be a cognitive process in the sense that it is a collective sense-making. A group of people are talking about a rumor to try to understand it. Studies of rumoring behavior during crisis events have a long tradition in classical studies of disaster by (Shibutani, 1966). More recently, the theory has been applied and extended to online contexts (Bordia & Difonzo, 2004; DiFonzo, 2009; Krafft et al., 2017). Rumors can also be a social process, where it serves as a conversational facilitator. People can be talking about a rumor to begin a conversation. Lastly, it can be an emotional process (Stieglitz & Dang-Xuan, 2013; Zeng & Zhu, 2019). Some rumors correlated with anxiety and relief. Finding out that they are not true, for instance, makes people relieved.

Although rumor spreading shows great resemblance to epidemics, rumor spreading is significantly different from epidemic spreading as individuals can decide to spread rumor whether or not. In the process of dissemination of rumor people will judge the authenticity of information through their cognition after getting the information.

Rumors are often discussed with hoax, misinformation, disinformation, and fake news. However, there are an important distinction between all of them. Hoax is deliberately fabricated falsehood made to masquerade as truth, misinformation is an honest mistake, and disinformation is a deliberate lie to mislead.



2. Data and Methodologies

2.1 Research design

Addressing the question 'How do animal rumors spread throughout Twitter during the COVID-19 information crisis?', this paper is a case study based on a subset of a collected Twitter dataset. Using visualization within a social network analysis study to track information propagation and information flow patterns, the analysis will contribute to understanding the nuances of rumor through highlighting the explanatory power of a graph tree. Comparing five rumor samples featuring animal in approaches provides breadth and depth on the subject, but also provides a stronger understanding of how dynamic attributes and interactions play roles in rumor diffusion through online social networks.

2.2 Case selection

The chosen cases are five rumors featuring animals that went viral on Twitter. These rumors are highly visible particularly in the dataset presented in this study.

With that being said, there are several reasons why rumor related to animals is important: (a) it represents variation to claims on COVID-19 infodemic, which contribute to divergent, (b) often considered as harmless that people are more willing to share it without thinking about its accuracy, not realizing that sharing causes harm too, (c) has the potential to penetrate deeper into the social network and propagate fast.

2.3 Data collection

Twitter provides free access to samples of public tweets which are posted on the platform through its Streaming API (Application Program Interfaces). On average around 6,000 tweets are being tweeted every second, which corresponds to over 350,000 tweets sent per minute, 500 million tweets per day, or around 200 billion tweet per year (Internet Live Stats, 2020).

With every tweet, Twitter provides multimodal data containing text, images, and videos, along with contextual and social metadata such as temporal and spatial information, and information about user connectivity and interactions. This robust user-generated data plays a significant role in making sense of user behavior and opinions.

To examine the differences in the information propagation patterns in rumors featuring animals, the study collected Twitter ID dataset from (Chen et al., 2020). In the dataset, there were 31,433,992 tweets over the course of a month after WHO declared a pandemic. In the preliminary analysis of this data, the study examined the list of most retweeted tweets to reveals certain topic grouping. Once the rumor was classified, the dataset contained 42,573 unique tweets generated by 39,758 unique users. The metadata information was collected with Hydrator, a tool to rehydrate Twitter IDs into full tweets. This is not the largest dataset, but in some ways it is sufficient to mine a partern.

Table 1. Animal related tweets. Source: Twitter

Animal related tweets	Controversial tweets
@Evanmcmurry: oh my god, the chicago aquarium closed due to coronavirus, so they let the penguins run around and check out the other exhibits. (staff was present.)	@ikaveri: Here's an unexpected side effect of the pandemic - the water's flowing through the canals of Venice is clear for the very first time in forever. The fish are visible, the swans returned.
@P_evans: I have seen a lot of random coronavirus-related stories but I did not see "monkey gang war" coming.	@RexChapman: Sandra the orangutang started washing her hands because she saw all the zookeepers doing it repeatedly during the COVID-19 crisis.
@_Jackhy: if the coronavirus has got you stressed out, monterey aquarium have closed to the public and are live streaming instead; you can see birds, sharks, otters, jellyfish, penguins, turtles, and plenty more - mbayaq.co/39xv9ny	@diageoliam: The World Health Organization has announced that dogs cannot contract Covid-19. Dogs previously held in quarantine can now be released. To be clear, WHO let the dogs out.
@Finnaganmarina: Shenzhen bans consumption of cats dogs snakes turtles frogs in attempt to stop spread if coronavirus.	@Spilling_The_T: While humans carry out social distancing, a group of 14 elephants broke into a village in Yunan province, looking for corn and other food. They ended up drinking 30kg of corn wine and got so drunk that they fell asleep in a nearby tea garden.
@Asiachloebrown: Can animals like dogs and cats get Covid-19?	@natureiscruel: Spotted Malabar civet... A critically endangered mammal not seen until 1990 resurfaces for the first time in India during lockdown.

Table 2 Annotated rumors and their tweet data summary. Source: Author's construction

Topic	Spreaders (Audience)	Retweet	Description (Example tweet)
Dog	14,944 (1,7E+07)	2,01E+09	WHO let the dogs out "The World Health Organization has announced that dogs cannot contract Covid-19. Dogs previously held in quarantine can now be released. To be clear, WHO let the dogs out."
Orangutan	4,228 (5,408,079)	4E+08	Orangutan washing hands during the pandemic "Sandra the orangutang started washing her hands because she saw all the zookeepers doing it repeatedly during the COVID-19 crisis. Wash your hands. Be more like Sandra."
Swan	12,239	4E+09	Swans returned to the Venice canals "Here's an unexpected side effect of the pandemic - the water's flowing through the canals of Venice is clear for the first time in forever. The fish are visible, the swans returned."
Elephant	9,728	7E+07	Drunk elephants in a tea garden "While humans carry out social distancing, a group of 14 elephants broke into a village in Yunan province, looking for corn wine and got so drunk that they fell asleep in a nearby tea garden."
Civet	1,434 (1,443,153)	6E+07	Rare Malabar civet spotted in India "Spotted Malabar civet... A critically endangered mammal not seen until 1990 resurfaces for the first time in India during lockdown."



Figure 1. Extracted time series, with x-axis as days and y-axis as the number of retweets on the topics (log). Source: Author's construction

(a) dog	(b) orangutan	(c) swan
(d) elephant * by hour	(e) civet

Figure 2. Followers compared to the number of retweets. Most users have followers between 0 to 100,000. Followers (x), number of retweets (y)



2.4 Analytical strategy

This study uses a social network analysis approach to visualize information flow in events of crisis, especially over Twitter, to analytically tracking the contribution of all users and exposing patterns that link network structure and dynamic information flow for rumors featuring animals. In the constructed network, nodes (otherwise called vertices) represent Twitter users, and connections (otherwise called edges) represent the retweet relation.

Retweet or 'RT' is a form of interaction most characteristic of Twitter as an information network. When a user retweets a tweet, they duplicate a piece of information they find interesting or agreeable posted by other users. By doing so, users that follow her/him are exposed to that information. This way, retweets are propagation mechanism that helps a piece of information spreads. Retweet graph has been well recognized as a reliable way to evaluate the information propagation and diffusion in Twitter (Cherepnalkoski & Mozetiи, 2016; Smith et al., 2014). A directed network draws an edge between two users (nodes) --- A and B -- if A retweets B. Link direction follows information flow. This research uses Gephi to visualize retweet networks. It is an open-source software for visualizing and analyzing large network graphs.

Each case will be explored individually through its graph, where in each graph the initial propagation pattern can be describes through conversational topology, subcommunities' identification through modularity calculation, and spotting the user that drives the conversation through eigenvector centrality. The tweet such users creates is often the most popular and widely repeated in the network. rumor quarantine strategy anxiety

2.5 Delimitations

The study has chosen to limit the tweets collected to only those written in English, since English was the common language among participating annotators.

Background

One of the things that makes COVID-19 pandemic different from predecessors is the dominance of social media in today's world. In the pre-internet era, information was curated by editorial gatekeepers and official government sources. But now, sense-making involves trying to gather information on the internet, through search engines and social media

One of the factors that also driving that is the way almost all news organizations now rely on social media (Shearer, 2018). For many readers, instead of clicking through to the actual reporting, the social media recap seems to be good enough. Too often social media is also allowing an agenda to be promoted, rather than news to be reported.

With that being said, navigating the information can be a challenge especially at the time of COVID-19 crisis, when the information flow is massive. Some of the information may be reliable, but a lot of it will not. Bad actors are manipulating social platforms for economic gain or other purposes. People retweet links without having looked at the site. And even innocently conceived jokes that can trigger panic. Over time and immediate, this may have detrimental effects, on individual and collective responses to the crisis event (Starbird, 2020).

This is also what led us to the phenomenon of viral untruths featuring animals. Since before the pandemic, people's appetite for animal content in social media is voracious (Molloy, 2011). With much of the world driven indoors to quarantine, some happy animal stories are flourishing on Twitter to soften the news about COVID-19 case surges, quarantine orders, and medical supply shortages.

Species not often see, or rarely in such large numbers, are exploring cities. Photographers and social media users have been duly documenting. There are goats along the streets of Welsh, flamingo flock to a locked-down Mumbai, Japanese sika deer walking the streets of Nara, dolphins in Venice canals, turkeys in Oakland, California, and news that ducks have returned to the fountains of Rome.

While these animal stories are popular, they got hundreds of thousands of retweets and went viral, many of these optimistic posts have turned out to be fake. There were no dolphins in Venice canals or drunken elephants ambling through China's Yunnan province. The narrative that wildlife populations will dramatically rebound and retake cities is fantasy - albeit one that might comfort those looking for meaning amidst the crisis.

As (Newman et al., 2012) put it: "When thoughts flow smoothly, people nod along." For similar reasons the more often a viral tweet appears in a people news feed, the more likely they think that it's true - even if they were originally skeptical. The trick has long been known by propagandists and peddlers of misinformation, but today's social media may exaggerate people's gullible tendencies.

Recent evidence shows that many people reflexively share content without thinking about their accuracy (Pennycook, McPhetres, et al., 2020). It suggests that people were sharing material that they could have known was false if they had thought about it more directly. It is because social media does not incentivize the truth. What it incentivizes is engagement (Pennycook, McPhetres, et al., 2020).



3. Findings

Figure 3. Network graphs of rumor spreading and distribution of retweets related to rumors on (a) dog ;(b) orangutan ;(c) swan ;(d) elephant; (e) civet. Each community is differentiated with color, each node sized by in-degree or the size of the audience. Source: Author's construction.

3.1 Rumor: WHO let the dogs out

There have been a lot of rumors floating around the topic of human-to-animal transmission. Panic began after a dog in Hong Kong was tested positive for the coronavirus during quarantine, then died three days after returning home (Reuters, 2020). Many pet owners posed questions to experts and the World Health Organization if their pets are at risk of contracting and spreading the disease. Thus, the rumor is closely tied to the sense-making activities of the crowd as it attempted to process official information about this topic.

Here is a tweet by @DiageoLiam, posted on March 12 at 11:36: "The World Health Organization has announced that dogs cannot contract Covid-19. Dogs previously held in quarantine can now be released. To be clear, WHO let the dogs out."

Figure 4. A tweet that made a pun after Baha Men hit song. Source: Author's screen capture, May 6,2020.

The tweet began as a pun. The pun was a copypasta, in which a copied and pasted block of text is posted on a message board instead of retweeting, which is more common. Copypasta could make the post look original, thus increasing the chance of people trusting its content. Five hours later, the tweet started getting a whole lot more engagement after a dog enthusiast-cum-television personality Gillian Turner @GillianHTurner reposted the tweet that reads: "BREAKING: The World Health Organization has announced that dogs cannot contract Covid-19. Dogs previously held in quarantine can now be released. To be clear, WHO let the dogs out."

'Breaking' is a term that broadcasters interchangeably used with breaking news to warrants an interruption of scheduled programming and or current news to report its details. Its use is also assigned to the most significant story of the moment or story that is being covered live. Many times, breaking news is used after the news organization has already reported the story. When a story has not been reported previously, the phrase 'just in' is sometimes used instead. In social media, it is a phrase used to describe when a story, meme, video, goes viral.

With the help of Turner, @DiageoLiam tweet reaching a new, larger, group of people. The statistics lend evidence that the average audience of this rumor before Turner repost it is 603 users. After she reposted it, the combine average audience rose to 1,466 users. Combine retweets from @DiageoLiam and Turner reached a peak volume of 2.5 million retweets per minute on March 12 at 18:00. The next day, the tweet circulates among tweets that posted about human-to-animal transmission.

Before the pun went viral, the World Health Organization firmly stated that there are no evidence that companion animals can be infected with the new coronavirus. After the pun spreads, and some people misinterpreted the joke, the organization removed a content tied to this topic on its myth-buster page.

This long-standing rumor was circulating for three weeks. The network consists of nine groups that formed a community clusters network type (see Figure 3 for visualization). This type of network group is formed around multiple news sources and popular topics, in this case human-to-animal transmission. The user in the center of this network is @DiageoLiam. Other users still retweet his tweet after 19 days.

To further assess differences in the unique rumor behaviors, the study selecting 100 random distinct tweets from communities and process it with AntConc. A program for analyzing electronic text (corpus linguistics) to find and reveal patterns in language. The most common words used in this rumor are dog, cat, coronavirus, animal, covid, human, people, pet, positive, tiger, and joke.

3.2 Rumor: Orangutan washing hands during the pandemic

The global situation due to the coronavirus has led to an increase in hygiene habits. This rumor correlates with a video of Sandra, an orangutan in Florida, washing hands. On April 1, a tweet by NBA player and social media star @RexChapman became viral. He captioned the video: "Sandra the orangutang started washing her hands because she saw all the zookeepers doing it repeatedly during the COVID-19 crisis. Wash your hands. Be more like Sandra."



Figure 6. A tweet claiming orangutan washing hands due to pandemic. Source: Author's screen capture, May 6, 2020.

Chapman's tweet becomes viral in less than one hour, but the claim was not true. The video was originally published by the Center for Great Apes back in November 2019 (Daly, 2020). While Sandra did not learn to wash her hands by copying zookeepers during the COVID-19 pandemic, the institution has used this viral video of Sandra to encourage people to follow health officials' guidelines and wash their hands for at least 20 seconds.

The graph resembles an in-hub and spoke network type called broadcast network. This type is often triggered by famous individuals who have loyal followers who retweet them. The community is star-as little interaction exists among members of the audience. Chapman have more than 700 thousand followers. Many followers come to his account for 'fun, feel-good Twitter content' (ESPN, 2020). In an hour, the tweet gains 24 million retweets. In a day, it gains 361 million retweets. The tweet got retweeted for eight consecutive days.

To further assess differences in the unique rumor behaviors, the study selecting 100 random distinct tweets from the thread and process it with AntConc. The most common words are: sandra, smart, human, orangutan, animal, like, monkey, many, washing, and hand.



Table 3. Content type vs period

rumor	content type	features	period (minute)	retweet count (minute)
swan	animals reclaiming urban spaces	Twitter collage + text	35,012	120746
dog	humor	text	31,408	63864
civet	animals reclaiming urban spaces	Twitter collage + text	17,224	3430
orangutan	humor	video url + text	12,330	25700
elephant	humor	embedded video + text	1,248	55364

3.3 Rumor 3: Swans return to Venetian canals

Amid the global lockdowns to curb the spread of the coronavirus, the social media platforms get flooded with images of wild animals who are reclaiming cities and streets around the world. On March, a Twitter user called @ikaveri shared photographs of Venetian swans with caption: "Here's an unexpected side effect of the pandemic - the water's flowing through the canals of Venice is clear for the first time in forever. The fish are visible, the swans returned."

In less than an hour the tweet garnered over a million retweets, two billion in one day. The tweet was later debunked by (National Geographic, 2020), that pointed out that swans are a familiar sight in the water of Burano, a small island in the greater Venice metropolitan area.

Twitter users began to introduce variations by restating the images in different words on March 17. It led to subsequent increases in propagation, from an average of 7,009 retweets per minute to 5 million retweets per minute on April 1. In the dataset, total retweet count for these images reached 442 million.

The swan rumor is a long-standing rumor that circulates for 24 days, 7 hours, and 32 minutes. There are three communities in this network, with @Shameka_xox0 as the most influential user followed by @ikaveri the author. The network graph formed a unified type that captures tight crowds or close communities where participants strongly connect to one another for information, ideas, and opinions.

Figure 7. A tweet claiming that swans return to Venetian canals. Source: Author's screen capture, May 6, 2020.

On Twitter, the author @ikaveri who lives in New Delhi, India, responded that she saw some photos on social media and decided to put them together in a tweet, unaware that the swans were already regulars in that particular place, even before the coronavirus tore across Italy. She also stated the unprecedented number of likes and retweets she is received on the tweet. It is a personal record and she would not like to delete it.

Out of context photos are very common source of misinformation. This form of misinformation can be particularly dangerous because images are powerful tool for swaying popular opinion and promoting false beliefs. Psychological research has shown that people are more likely to believe true and false trivia statements when they are presented alongside an image (Newman et al., 2012). In addition, pictures can alter what people remember.

To further assess differences in the unique rumor behaviors, the study selecting 100 random distinct tweets from communities and process it with AntConc. The most common words are people, face, human, swan, water, world, earth, nature, see, venice.

3.4 Rumor 4: Drunk elephants in a tea garden

This is a short-lived rumor that circulates only for 20 hours and 48 minutes. The rumor started on 19 March when a user called @Spilling_The_T shared a photograph of sleeping elephants passed out after drinking corn wine in a village in Yunnan Province, as the caption claimed. The post reads: "While humans carry out social distancing, a group of 14 elephants broke into a village in Yunan province, looking for corn and other food. They ended up drinking 30kg of corn wine and got so drunk that they fell asleep in a nearby tea garden."

Figure 8. A tweet claiming elephants passed out after drinking corn wine. Source: Author's screen captures on May 6, 2020.

Photos of elephants raiding a remote village in Yunnan province and getting drunk first surfaced on Weibo. The Chinese microblogging platform, Chinese news outlet Xinhua News reported on 16 March. The reality is a group of 14 elephants entered several villages in Yunnan province's Menghai County on March 11, damaging roofs and overturning wine jars, according to official reports. However, it was difficult to confirm if the elephants drank any of the alcohol that they spilled, according to Xinhua News.

Five other accounts shared the tweet with slightly different variations, and when accumulated, the topic has been retweeted for approximately 1.7 million times. In the dataset, the rumor lasted for about nine hours. It reached a peak volume of 2 tweets and 116,223 retweets per minute at about 15:00 March 19. By the end of the day, the total retweet count for this tweet reached 69 million.

The graph also formed a unified network type that captures tight crowds. There are four communities in this network, with account called @Spilling_The_T in the center or the most influential.

3.5 Rumor 5: Rare Malabar civet spotted in India

On 26 March, a clip showing a civet started circulating on social media. The video was reportedly taken in Meppayur, a town in the Kozhikode district of the Kerala state in southern India. The video is real, filmed when many people were practicing social distancing to help slow the spread of COVID-19. Hence, the video was widely shared along with a misleading caption: "Spotted Malabar civet... A critically endangered mammal not seen until 1990 resurfaces for the first time in India during lockdown."



Figure 9. A tweet claiming that a critically endangered civet resurfaces in India during lockdown. Source: Author's screen capture on May 6, 2020.

The civet in the video is not the critically endangered Malabar civet, but more likely a common small Indian civet. Parveen Kaswan, a member of the India Forest Service, identified the animal as a small Indian civet on Twitter. Kalyan Varma, a wildlife photographer also identified it as the small Indian civet, not the Malabar civet. Conservation India writes that the small Indian civet and the Malabar civet are similar in their appearance and both have a banded tail and similar markings on the bodies.

Although the video likely shows the relatively common small Indian civet, still it is an unusual sight given it is a nocturnal animal. Some social media users suggested that the animal could be sick as it appears to be moving a bit awkwardly and slow. This type of rumor is pretty common during the COVID-19 pandemic. For the most part, the rumors have been exaggerated like swans returning to the waterways of Italy, orangutan washing hands, and the drunk elephants.

The graph looks similar to the orangutan rumor, it is a broadcast network. All nodes have access to a central node, a user called @NatureisCruel. There are 1,410 in-degree or users that retweeted the message.

In the dataset, the rumor sustains for 12 days. It reached a volume of 2 tweets and 66,614 retweets per minute at about 15:00 March 19. The total retweet count for this tweet reached 59,086,758.

Table 4. Users topping the list of retweet counts. Their influence in the network calculated with Pearson's Correlation Coefficient. Source: Author's calculation

	Popular Accounts	r
1	Shameka_xox0 (N = 7030)	0.012
2	DiageoLiam (N = 5631)	0.019
3	Spilling_The_T (N= 5514)	-0.003
4	Ikaveri (N = 4868)	0.002
5	GillianHTurner (N = 3044)	0.018
6	RexChapman (N = 4228)	-0.004
7	LiquidFaerie (N = 2174)	0.052
8	BornWitaCharm (N = 1257)	-0.032
9	NatureisCruel (N = 1434)	0.009
10	MeetTheWoo2 (N = 711)	-0.004

More findings

The findings of the study revealed that:

(i) rumors featuring animals show quite a remarkable size of audience on Twitter. In this study, the smallest animal rumor reaches one million users, with the largest reach nearly 17 million users.

(ii) rumors featuring animals propagate fast. In this study, every sample show that they reach more than one million retweets in less than one hour after being posted. The average elapsed time for a retweet is 1 minute and 28 seconds.

(iii) rumors that circulate in a clustered network types tend to penetrate deeper to many small and medium groups, for this reason, they continue to earn retweets and have a much longer lifespan on the network.

(iv) networks with shorter paths, like the in-hub and spoke networks related to an orangutan washing hands and the one related to the civet, follow a more constant rate of propagation. While networks with relatively longer paths, like the rumor related to the dog, swan, and elephant, immediately peak. In some cases, there are multiple peaks.

(v) networks with relatively long path are more conversational, they earn more retweets per minute compared to in-hub and spoke networks.

(vi) authors expand their audience base through the viral propagation of animal rumors. The samples in this study show that some people who spread the rumors had fewer followers, followed fewer people, and tweeted less often.

(vii) the type of people who start and spread the viral animal rumors are dominated by jokers, who claim they seek to lighten the mood by sharing joyful and happy post about animals.

(viii) Some authors in the sample admitted they are getting a lot of attention online from sharing posts related to animals. In social media, an immediate social reward gives one's self-esteem a temporary boost (Gentile et al., 2012). Users who got a lot of attention online from the unprecedented number of likes and retweets have refused to remove their posts even after debunked for sharing misleading and untruths. (ix) the network visualization highlights how the untruths could be penetrated deeper and wider. In @DiageoLiam cases, from the graph network, we could see that she got connected to @GillianHTurner through more than one bridge or information broker. The author discovered that even after the peak of retweeting has passed, @DiageoLiam's tweet could resurface again when the general animal-to-human topic was discussed on Twitter in a different community. The presence of bridges lets the rumor to penetrate deeper into the network.

That bridges gave @DiageoLiam the magnitudes to reach a larger audience and connected to the @GillianHTurner audience. In this case, the retweet network indicated that the number of followers on Twitter does not always equal influence. Instead, influential users in a network are more important to making connections with other users and to control the flow of information.

(x) the content involves various forms of reconfiguration, where existing information is spun, twisted, recontextualized. Hence, there is no examples of deep fakes.



Conclusion

This dissertation has sought to examine how rumors are shaped in the time of information crisis, through highlighting the relationship between information and its flow patterns in the Twitter social platform. Through visualizing and investigating retweet networks of five viral rumor featuring animals, it has demonstrated how certain rumor can propagate faster and penetrates deeper in the network, by utilizing the retweet functionality support via Twitter. By posting a particular content, a user is like being able to jump into a networking function with hundreds of people, and through a retweet, it could propagate faster.

A time of crisis like COVID-19 pandemic can also be a prime lying season for people who want attention. Twitter, in particular, makes it easy to embed a couple of photos, claim something that is not happening is happening, and watch tens of thousands of people online validate the author for it.

Much of what the study learn from the samples is some users actively looking for persuasive content on the internet, reposted it as a copypasta to make it look original or simply arrange it into interesting photo collage to make it believable enough and interesting enough to be consumed through microblogging like Twitter. The information flow tends to be either satirical or informative but exaggerated and sensational. Many people, if not everyone, on Twitter has fallen prey to this at some point because retweeting is faster and easier than googling something.

Fake feel good stories can be seen as harmless. Hence, they diffused father, faster, deeper, and more broadly than the truth in all categories of information. This supported previous findings on the rapidity of falsity on Twitter (Vosoughi et al., 2018).

Ultimately, the pattern of retweet network in animal rumors highlights how information flows form cascades with different profiles: for examples a fast-spreading rumor that is quickly snuffed out would have high breadth but little depth and low virality.

At certain rate, reduce the spread of rumor through press conference or official statements would be less effective. Hence, this would be valuable for further research. Through the network analysis, the study suggests that human plays significant roles for the spread by sharing without thinking.

To help increase the collective health of public conversation, the more effective way to slowing the propagation of viral untruths - which is just another kind of virus - is by stop sharing it. The retweet patterns demonstrate how significance the number of propagations could happen in a day, and that particular tweet kept being retweeted in long term.



Bibliography

1. Mintz, A. P. (2012). Web of deceit : misinformation and manipulation in the age of social media. CyberAge Books.

2. Nielsen, R. K., Fletcher, R., Newman, N., Brennen, J. S., & Howard, P. N. (2020). Navigating the ` Infodemic ': How People in Six Countries Access and Rate News and Information about Coronavirus (Issue April).

3. Stieglitz, S., & Dang-Xuan, L. (2013). Emotions and information diffusion in social media - Sentiment of microblogs and sharing behavior. Journal of Management Information Systems, 29(4), 217-248.

4. Vosoughi, S., Roy, D., & Aral, S. (2018). The spread of true and false news online. Science, 359(6380), 1146-1151.

5. Wooley and Howard. (2019). Computational propaganda: political parties, politicians, and political manipulation on social media. In S. C. and P. N. H. Wooley (Ed.), International Affairs. Oxford University Press.

Размещено на Allbest.ru

...