New research analyzes millions of twitter posts during hurricanes to understand how people communicate in a disaster
This study suggests that monitoring social media during hurricanes could help communities better plan for and mitigate the impacts of climate change
Peer-Reviewed PublicationIn the face of a potentially disastrous storm like Hurricane Ida, people take to Twitter and other social media sites to communicate vital information. New research published in the journal Risk Analysis suggests that monitoring and analyzing this social media “chatter” during a natural disaster could help decision makers learn how to plan for and mitigate the impacts of severe weather events in their communities.
Jose E. Ramirez-Marquez from the Stevens Institute of Technology and Gabriela Gongora-Svartzman from Carnegie Mellon University's Heinz College performed an analysis of more than six million Twitter posts over time during three major hurricanes that made landfall in 2017: Harvey (Texas), Irma (Florida), and Maria (Puerto Rico). The goal of their study was to develop and test a new method for measuring social cohesion, an important factor in a community’s resilience during the severe weather events brought on by climate change.
The methodology presented in Risk Analysis involves combining and implementing text processing techniques and graph network analysis to understand the relationships between nine different categories of Twitter users during a hurricane. These include citizens, media, government, entertainment, business, charity-NGOs-volunteers, sports, technology-science-education, and other verified accounts. Knowing who the participants are behind the messages can help researchers identify how authorities communicate which kinds of messages, how people affected by the hurricanes interact with them, and what their needs are.
Visualizations incorporated into the study illustrate the connections between social media participants and the degree of social cohesion throughout each hurricane’s timeline.
Social cohesion has been described as “the glue that holds society together.” It affects how a community comes together in times of need. Social cohesion can help reduce the number of vulnerabilities experienced by a community during a disaster and reduce the time it takes to rebuild. The stronger the social cohesion, the more resilient a community is.
“If we measure and understand social cohesion within communities, then we can seek to increase cohesion - through policies, community programs, and other strategies - which in turn will increase the resilience of communities,” says Gongora-Svartzman. “People in a more resilient community are more willing to volunteer and help each other during a disaster. They are also better informed, therefore knowing who to ask for help, what resources are available, and how to help during a disaster.”
Visualizations in the study illustrate the seven metrics that are combined to create a single measurement of social cohesion. One of those metrics is information dissemination. This refers to the intensity of tweets, or communication between participants, during the timeline captured for each hurricane. This timeline of social media activity for each hurricane shows how active participants were on each day before, during, and after the hurricane. A graph of the data shows that the intensity of communication peaks for each hurricane shortly before or after it makes landfall. In the case of Maria in Puerto Rico, the analysis shows that a significant amount of conversation continues for more than a week after the hurricane ends – signifying that post-disaster management strategies were being put in place, rescues were occurring, and rebuilding efforts were starting to evolve.
The researchers hope this new method for tracking and visualizing social media communications during a severe storm can contribute to future risk management and disaster mitigation policies. “Because we identify the types of actors in a social network and how this network varies daily, decision makers could use this measurement to release strategic communication before, during, and after a disaster strikes – thus providing relevant information to people in need,” says Ramirez-Marquez.
In light of the disastrous impacts of Hurricane Ida on the people of New Orleans, he adds, it is important to understand what happened during each storm to mitigate the impacts on the most vulnerable people. “If we had a national database of the social media communications pre-during-post disaster then we would be able to better identify the needs of a community and the limitations of current policy and response,” says Ramirez-Marquez. “It is concerning that the communities that experienced the harshest effects during Katrina will again be harshly affected during Ida. This shows a lack of learning from past events.”
###
About SRA
The Society for Risk Analysis is a multidisciplinary, interdisciplinary, scholarly, international society that provides an open forum for all those interested in risk analysis. SRA was established in 1980 and has published Risk Analysis: An International Journal, the leading scholarly journal in the field, continuously since 1981. For more information, visit www.sra.org.
JOURNAL
Risk Analysis
ARTICLE TITLE
Social Cohesion: Mitigating Societal Risk in Case Studies of Digital Media in Hurricanes Harvey, Irma, and Maria
ARTICLE PUBLICATION DATE
8-Sep-2021
What AI analysis of 100 million social media interactions can teach product managers
News from the Journal of Marketing
Peer-Reviewed PublicationIMAGE: BRAND RELATIONSHIP CHART view more
CREDIT: YI YANG, KUNPENG ZHANG, P.K. KANNAN
Researchers from Hong Kong University of Science and Technology and University of Maryland at College Park published a new paper in the Journal of Marketing that examines product-markets from the perspective of consumers’ perceptions of brands instead of from purchase data.
The study, forthcoming in the Journal of Marketing, is titled “Identifying Market Structure: A Deep Network Representation Learning of Social Engagement” and is authored by Yi Yang, Kunpeng Zhang, and P. K. Kannan.
Firms compete to satisfy consumers’ specific needs. The market and the competing products comprise a “product-market” with a boundary. Identifying the product-market boundary and examining the strength of competition among brands within the product-market have long been important issues for managers. It has implications for product design, product positioning, new customer acquisition, and pricing and promotion decisions. Rapid changes to the competitive environment, however, have made identifying product-market boundaries increasingly challenging. The traditionally defined SIC and NAICS classification codes may not be adequate, and especially not for capturing consumers’ perceptions of and preferences for brands.
Technological advances change product-market boundaries. For example, film cameras gave way to digital cameras, then the digital camera product-market was upended by technological developments in smartphones. Similarly, Ford recently introduced its F-150 Lightening electric pick-up truck at a low price of $40,000 to remove a major barrier for customers thinking about making a switch from gasoline engines and Tesla introduced its electric Model 3 starting of $39,500 to broaden its appeal to mass-market car buyers. Both moves thereby changed competition within these lower-end vehicle product-markets. Companies also increasingly enter product-markets they previously did not compete in. For example, Amazon, hitherto an online platform, essentially crossed product-market boundaries when it acquired Whole Foods, thus presenting traditional grocery brands with a new and innovative competitor. Similarly, Whirlpool Corporation, the world’s largest home appliance maker, acquired Yummly, a recipe search engine with 20 million users, getting itself closer to how its potential consumers cook.
The reality is that produce-markets are more fluid than ever. Given the potential for new and unforeseen relationships between brands, managers need deeper insights into the fluid product-market boundaries. How can managers accurately identify potential threats and opportunities, especially those in different product-markets? How can managers derive these insights using easy-to-obtain and publicly available data? A new study in the Journal of Marketing addresses these questions and derives marketing insights using big data (over a hundred million social media user engagement “likes” and “comments”) spanning several thousands of brands in different product/service categories. This research looks at product-markets from a different perspective by focusing on consumers’ perceptions of brands based on social media engagement data, which unveils more of the dynamics at play, rather than using purchase data that are locked within pre-specified product-market boundaries. “We can show that two brands are very close to each other, even though they are in completely different SIC categories. In other words, by examining brand-user relationships, we generate a more inclusive and current representation of brands and the competitive/complementary relationships among them,” says Yang.
Using brand engagement data involving millions of social media users, the researchers capture latent relationships among thousands of brands and across many categories to reveal a highly precise market structure. They build a brand-user network using the data and then compress the network into a market structure map that visually represents the brands. Readers can view and interact with the map at https://market-structure.github.io/index.html.
“For example,” explains Zhang, “consider Amazon’s acquisition of Whole Foods. In the chart below, the green lines show that Whole Foods’ proximity to other retail brands such as Target and Walmart increases, while the red lines shows its proximity to supermarket brands such as Goya Foods and HelloFresh decreases.”
Our market structure map helps managers identify brands outside of the product-market that are close to a specific brand. For example, Disney Cruise Line and Hyatt are two brands outside the airline product-market, but are identified as proximal brands to Southwest but not for United. Such findings can provide opportunities for Southwest to target users who like Disney Cruise and Hyatt in social media. Southwest can cross-promote with Disney Cruise and/or Hyatt on each other’s websites or launch coalition loyalty programs. From the viewpoint of other hotel chains competing with Hyatt, gleaning such insights early on may help them take proactive actions. The figure below depicts these relationships.
This research reveals that managers can obtain very useful insights from user engagement data on social media platforms at a scale and scope that cannot be easily matched by any other source. Kannan says “The power of our method lies in its ability to capture the dynamic changes in market structure. Since the maps are based on the analysis of big data that can be collected in a relatively short time, our methodology can track changes in brands’ relative position when firms introduce new products, new promotions, and new marketing initiatives. Firms can deploy our method to enhance their social network-based marketing efforts by better targeting specific potential customers.”
Full article and author contact information available at: https://doi.org/10.1177/00222429211033585
About the Journal of Marketing
The Journal of Marketing develops and disseminates knowledge about real-world marketing questions useful to scholars, educators, managers, policy makers, consumers, and other societal stakeholders around the world. Published by the American Marketing Association since its founding in 1936, JM has played a significant role in shaping the content and boundaries of the marketing discipline. Christine Moorman (T. Austin Finch, Sr. Professor of Business Administration at the Fuqua School of Business, Duke University) serves as the current Editor in Chief.
https://www.ama.org/jm
About the American Marketing Association (AMA)
As the largest chapter-based marketing association in the world, the AMA is trusted by marketing and sales professionals to help them discover what is coming next in the industry. The AMA has a community of local chapters in more than 70 cities and 350 college campuses throughout North America. The AMA is home to award-winning content, PCM® professional certification, premiere academic journals, and industry-leading training events and conferences.
https://www.ama.org
JOURNAL
Journal of Marketing
ARTICLE TITLE
Identifying Market Structure: A Deep Network Representation Learning of Social Engagement
ARTICLE PUBLICATION DATE
8-Jul-2021
Video fake news believed more, shared more than text and audio versions
UNIVERSITY PARK, Pa. — People are more likely to believe fake news in a video format compared to text and audio forms of the same story, according to a team of researchers. They added that people are also more willing to share these videos with people in their network.
In a study, about 58% of people who viewed a fake news video on an instant messaging smartphone app believed the video was real, compared to 48% of the people who heard the same story in audio format. Only 33% of the audience who read the article found the information credible.
The researchers said that the impulse to “believe what you see” may make video fake news a more stealthy and possibly more dangerous form of social media manipulation.
“When you see something, you believe in it more because it seems so obviously real,” said S. Shyam Sundar, James P. Jimirro Professor of Media Effects in the Donald P. Bellisario College of Communications, co-director of the Media Effects Research Laboratory and an affiliate of the Institute for Computational and Data Sciences. “With text, you must use your imagination and put yourself in that situation or scene that's described in a series of words. But with video, it's a much more direct experience. You are seeing it and you're feeling it. And so, people seldom stop to think when they see a video, that they've seen something that is not true.”
The study shows that this tendency to believe more in video fakes is higher among people who are less involved in the topic of the story, whereas those who are highly involved tend to be persuaded more by text and audio fakes, said the researchers.
Viral videos
People were also more likely to share the video compared to audio- or text-based versions of the story, according to Maria D. Molina, assistant professor of advertising and public relations, Michigan State, who worked with Sundar on the study. About 78% of study participants who saw the video said they would share the video with others in their network, compared to 63% who heard the story and 67% who read the story.
The researchers, who reported their findings in the Journal of Computer-Mediated Communication on Aug. 1, said the use of artificial intelligence to create highly realistic, video fake news — or deep fakes — make the findings even more worrisome.
“There’s a powerful psychological appeal behind video form,” said Sundar. “And recent technologies can be used to very easily manipulate videos to create deep fakes. If people are going to believe what they see with their own eyes, then, these deep fakes have potentially dangerous consequences.”
The researchers studied the reaction to videos on WhatsApp, the largest instant message platform in the world and very popular in the global south, according to the team. They exposed 180 participants from rural and urban areas of India to three fake news stories obtained from Boom Live, Alt News and Social Media Hoax Slayer, websites that specialize in exposing false online viral videos in India.
After answering demographic questions, participants were shown a WhatsApp message on the researcher’s phone with one of the three stories, in one of the three formats — video, audio, or text — and then asked to imagine that they came across this story in one of their WhatsApp groups.
After they reviewed the story, participants answered a series of questions about the story.
The researchers decided not to share the fake news stories directly with the participants’ smartphones. Molina said there was a risk that the experimental fake news stories could be spread by participants.
"We are actually showing them the stories from another phone, but we can envision the effects that we found being even stronger if people actually receive these messages in their own WhatsApp feed,” said Molina.
Serious consequences
Sundar said the study was motivated by recent mob riots and lynchings in India, which were triggered by fake videos spread through social media.
“A lot of us have been talking about fake news for a while and we think about it as spreading lies, but these incidents in India opened my eyes to the serious deadly consequences of fake news in video format,” said Sundar. “It seems to evoke a visceral reaction rather than a thoughtful response."
The video format is quite complex, according to the researchers, making it difficult to pay close attention to the accuracy of the information.
“Video contains many streams of information, including audio, visuals, moving images, graphics and text,” said Sundar. “When people take in all this information, they are mentally overloaded, making them less likely to scrutinize the details.” This, along with the seeming realism of what is portrayed, makes videos more likely to be believed and shared, he added.
WhatsApp messages are encrypted on both ends, which means the general public cannot see and correct fake information by commenting on the messages. Only the intended recipients can see the fake posts.
The researchers offered a few ways to combat possible damage caused by video-based fake news. For example, social media moderators may want to prioritize video posts when they review media that has been flagged as possible fake news.
The researchers also suggest making users aware of the power of video, but acknowledge that it may not be fully successful. While media literacy can be a component of efforts to curb the spread and power of video-based fake news, the power of the biases and mental shortcuts evoked by video will be difficult to limit just by informing the public, said Molina.
As a solution, the researchers suggest introducing interactive features that would give users pause and make them reflect on the post before rushing to forward it. They propose that platforms like WhatsApp use alerts and warnings, as and when users encounter videos.
Eugene Cho, assistant professor of communications studies, the College of New Jersey, worked with Sundar and Molina on this study.
WhatsApp supported the work through its Research Awards for Social Science and Misinformation. The staff of the Centre for Media Studies, New Delhi, India, assisted in study design and data collection.
###
JOURNAL
Journal of Computer-Mediated Communication
ARTICLE PUBLICATION DATE
1-Aug-2021
No comments:
Post a Comment