scholarly journals Healthy vs. Unhealthy Food Images: Image Classification of Twitter Images

Author(s):  
Tejaswini Oduru ◽  
Alexis Jordan ◽  
Albert Park

Obesity is a modern public health problem. Social media images can capture eating behavior and the potential implications to health, but research for identifying the healthiness level of the food image is relatively under-explored. This study presents a deep learning architecture that transfers features from a 152 residual layer network (ResNet) for predicting the level of healthiness of food images that were built using images from the Google images search engine gathered in 2020. Features learned from the ResNet 152 were transferred to a second network to train on the dataset. The trained SoftMax layer was stacked on top of the layers transferred from ResNet 152 to build our deep learning model. We then evaluate the performance of the model using Twitter images in order to better understand the generalizability of the methods. The results show that the model is able to predict the images into their respective classes, including Definitively Healthy, Healthy, Unhealthy and Definitively Unhealthy at an F1-score of 78.8%. This finding shows promising results for classifying social media images by healthiness, which could contribute to maintaining a balanced diet at the individual level and also understanding general food consumption trends of the public.

Author(s):  
Caroline Pauletto Spanhol Finocchio ◽  
Homero Dewes

Purpose – The purpose of this paper is to characterize food consumption outside the home and verify the existence of relationships between income, expenditure and the prevalence of overweight and obesity in adult individuals in Brazil. Design/methodology/approach – Food expenditure data were used as well as those on the prevalence of overweight and obesity available in the reports from Family Budgets Research carried out by the Brazilian Institute of Geography and Statistics (IBGE) in the years 2002-2003 and 2008-2009. In addition, the authors used the Kruskall-Wallis test to verify whether there were significant differences between the classes of income with respect to the average expenditure on food outside the home. Findings – It was observed that the higher the income of the individual, the greater the expenditure on food outside the home. It was also found that the prevalence of overweight and obesity is higher in individuals who have higher income, particularly among men. Thus, it is suggested that the higher the income of the individual, the larger the expenditure on food outside the home and the greater the prevalence of overweight and obesity. Research limitations/implications – This study used spending on food away from home (FAFH), but the authors know that is need to use other variables like frequency of FAFH and quantities but this data were not available. Practical implications – The study points to the importance of restaurants in the prevention of obesity, since they can offer in their menus foods with fewer calories, and they can also increase the variety and availability of fruits and vegetables. Originality/value – The study discusses the public health problem, obesity, at the same time as it presents the importance of agribusiness in providing a balanced diet for individuals.


Author(s):  
Trevor Diehl ◽  
Brigitte Huber ◽  
Homero Gil de Zúñiga ◽  
James Liu

Abstract This study explores the individual- and country-level factors that influence how getting news from social media relates to people’s beliefs about anthropogenic climate change. Concepts of psychological distance and motivated reasoning are tested using multilevel analysis with survey data in 20 countries (N = 18,785). Results suggest that using social media for news is associated with a decrease in climate skepticism across the sample. However, social context at the individual-level (conservative political ideology and low trust in science) and at the macro-level (high gross domestic product and individualism) moderate the effect, and therefore reduce social media’s potential to inform the public about climate change. This study contributes to conversations about the ability of emerging media to address science issues, particularly in developing countries.


2021 ◽  
pp. 103530462110555
Author(s):  
Sue Williamson ◽  
Linda Colley ◽  
Meraiah Foley

Before the COVID-19 pandemic forced large sections of the workforce to work from home, the uptake of working from home in the public sector had been limited and subject to the discretion or ‘allowance decisions’ of individual managers. Allowance decisions are influenced by factors at the organisational, group and individual levels. This research examines managers’ allowance decisions on working from home at each of these levels. It compares two qualitative datasets: one exploring managerial attitudes to working from home in 2018 and another dataset collected in mid-2020, as Australia transitioned out of the initial pandemic lockdown. The findings suggest a change in the factors influencing managers’ allowance decisions. We have identified a new factor at the organisational level, in the form of local organisational criteria. At the group level, previous concerns about employee productivity largely vanished, and managers experienced an epiphany that working from home could be productive. At the individual level, a new form of managerial discretion emerged as managers attempted to reassert authority over employees working remotely. These levels intersect, and we conclude that allowance decisions are fluid and not made solely by managers but are the result of the interactions between the organisational, group and individual levels. JEL Codes J81, J32


2018 ◽  
Vol 62 (6) ◽  
pp. 796-815 ◽  
Author(s):  
Lorenzo Zamponi ◽  
Lorenzo Bosi

Alternative action organizations (AAOs) are collective bodies engaged in carrying out alternatives to dominant socioeconomic and cultural practices through actions that aim to provide people with alternative ways of enduring day-to-day difficulties and challenges in hard economic times. They are often interpreted as merely “philanthropic” actors, although it is not rare to see them go beyond the provision of direct services to people in need and end up pursuing political goals through political means. This article focuses on the process of politicization, that is, the transition of issues from the private to the public sphere and thus the use of public forms of contention (e.g., protest) proposing public solutions at the collective level instead of private solutions at the individual level. We argue for the role of the crisis in the politicization of AAOs. In particular, we show that the appropriation of the context as a context of economic crisis in the discourse of AAOs has a visible effect on their politicization, in terms of both repertoire of actions and goals. Furthermore, we show that social solidarity organizations, those that are not inherently politicized, are the main protagonists of this crisis-triggered transition. The article draws on statistical analysis of the data collected through the coding of AAOs’ websites in Greece, Italy, and Spain.


2021 ◽  
Author(s):  
Gaurav Chachra ◽  
Qingkai Kong ◽  
Jim Huang ◽  
Srujay Korlakunta ◽  
Jennifer Grannen ◽  
...  

Abstract After significant earthquakes, we can see images posted on social media platforms by individuals and media agencies owing to the mass usage of smartphones these days. These images can be utilized to provide information about the shaking damage in the earthquake region both to the public and research community, and potentially to guide rescue work. This paper presents an automated way to extract the damaged building images after earthquakes from social media platforms such as Twitter and thus identify the particular user posts containing such images. Using transfer learning and ~6500 manually labelled images, we trained a deep learning model to recognize images with damaged buildings in the scene. The trained model achieved good performance when tested on newly acquired images of earthquakes at different locations and ran in near real-time on Twitter feed after the 2020 M7.0 earthquake in Turkey. Furthermore, to better understand how the model makes decisions, we also implemented the Grad-CAM method to visualize the important locations on the images that facilitate the decision.


MEDIASI ◽  
2021 ◽  
Vol 2 (2) ◽  
pp. 100-107
Author(s):  
Shania Shaufa ◽  
Thalitha Sacharissa Rosyidiani

This article explains about online media iNews.id in implementing gatekeeping function. This study aims to find out how gatekeeping efforts iNews.id in the production process on the issue of preaching restrictions on worship in mosques during Ramadan in 2020. During the Covid-19 pandemic, the current media situation, especially in the midst of a crisis, encourages the public to become heavily dependent on media coverage. With a qualitative approach, researchers analyzed five levels of influence on the gatekeeping process in online media iNews.id. The results of this study show that factors that influence the way iNews.id in the production process of preaching restrictions on worship in mosques due to the Covid-19 pandemic are the individual level of media workers, the level of media routine, the organizational level, the extramedia level, and the social system level. The conclusions of this study state the most dominant levels is the organization level and the media routine level in the iNews.id.


Author(s):  
Ron Astor ◽  
Rami Benbenishty

On their own, photos and videos are not a reliable source of information about what is taking place in a school. It’s easy to react emotionally or with outrage to a video of a fight, a child being picked on, or some other display of abuse or wrongdoing. But everyone knows from highly publicized incidents posted on YouTube, Twitter, or other social media sites that photos and videos can be taken out of context. They tell a story, but they don’t tell the whole story. Even so, they can be used by administrators to discern whether the action shown in the photo or video is an isolated incident or could be a symptom of a larger problem. If an alarming photo or video taken at a school is receiving attention from the media, it’s better to talk about it with the school community as soon as possible than to pretend it didn’t happen. Situations like these create an opportunity to examine and share other sources of data about school safety, violence, and victimization. Too often, one incident can cause the public to draw conclusions about a school that are not accurate. That’s why a monitoring system is necessary— to put such an incident in context. Administrators who can refer to other sources of data regarding violence, drug use, or weapons can respond with more confidence when faced with criticism over one incident. As part of a monitoring system, photos, videos and other technology can be used for positive purposes. They allow students who might skip questions on a survey or don’t want to speak up during a focus group to express themselves in a different way. There are many examples of projects in which students are given cameras and microphones and encouraged to express themselves and present their experiences in school through this media. In addition to the individual students benefitting from such experiences, school leaders, staff members, and parents get the opportunity to see the school from the students’ perspectives.


2020 ◽  
Vol 38 (5) ◽  
pp. 587-601
Author(s):  
Helen Cripps ◽  
Abhay Singh ◽  
Thomas Mejtoft ◽  
Jari Salo

PurposeThe purpose of this research is to investigate the use of Twitter in business as a medium for knowledge sharing and to crowdsource information to support innovation and enhance business relationships in the context of business-to-business (B2B) marketing.Design/methodology/approachThis study uses a combination of methodologies for gathering data in 52 face-to-face interviews across five countries and the downloaded posts from each of the interviewees' Twitter accounts. The tweets were analysed using structural topic modelling (STM), and then compared to the interview data. This method enabled triangulation between stated use of Twitter and respondent's actual tweets.FindingsThe research confirmed that individuals used Twitter as a source of information, ideas, promotion and innovation within their industry. Twitter facilitates building relevant business relationships through the exchange of new, expert and high-quality information within like-minded communities in real time, between companies and with their suppliers, customers and also their peers.Research limitations/implicationsAs this study covered five countries, further comparative research on the use of Twitter in the B2B context is called for. Further investigation of the formalisation of social media strategies and return on investment for social media marketing efforts is also warranted.Practical implicationsThis research highlights the business relationship building capacity of Twitter as it enables customer and peer conversations that eventually support the development of product and service innovations. Twitter has the capacity for marketers to inform and engage customers and peers in their networks on wider topics thereby building the brand of the individual users and their companies simultaneously.Originality/valueThis study focuses on interactions at the individual level illustrating that Twitter is used for both customer and peer interactions that can lead to the sourcing of ideas, knowledge and ultimately innovation. The study is novel in its methodological approach of combining structured interviews and text mining that found the topics of the interviewees' tweets aligned with their interview responses.


Gender Issues ◽  
2020 ◽  
Author(s):  
Zaida Orth ◽  
Michelle Andipatin ◽  
Brian van Wyk

Abstract Sexual assault on campuses has been identified as a pervasive public health problem. In April 2016, students across South African universities launched the #Endrapeculture campaign to express their frustration against university policies which served to perpetuate a rape culture. The use of hashtag activism during the protest served to spark online public debates and mobilize support for the protests. This article describes the public reactions to the South African #Endrapeculture protests on the Facebook social media platform. Data was collected through natural observations of comment threads on news articles and public posts on the student protests, and subjected to content analysis. The findings suggest that the #nakedprotest was successful in initiating public conversations concerning the issue of rape culture. However, the reactions towards the #nakedprotest were divided with some perpetuating a mainstream public discourse which perpetuates rape culture, and others (re)presenting a counter-public that challenged current dominant views about rape culture. Two related main themes emerged: Victim-blaming and Trivialising Rape Culture. Victim-blaming narratives emerged from the commenters and suggested that the protesters were increasing their chances of being sexually assaulted by marching topless. This discourse seems to perpetuate the notion of the aggressive male sexual desire and places the onus on women to protect themselves. Other commenters criticised the #nakedprotest method through demeaning comments which served to derail the conversation and trivialise the message behind the protest. The public reaction to the #nakedprotest demonstrated that rape culture is pervasive in society and continues to be re(produced) through discourse on social media platforms. However, social media also offers individuals the opportunity to draw from and participate in multiple counter-publics which challenge these mainstream rape culture discourses.


2015 ◽  
Vol 19 (3) ◽  
pp. 239-253 ◽  
Author(s):  
Jie Xu ◽  
Yiye Wu

Purpose – The purpose of this paper is to investigate the effects of using Twitter on American stakeholders’ crisis appraisal for organizations originated from two foreign countries with distinctively different perceptions. Design/methodology/approach – This study uses a 2 (medium: Twitter vs news release)×2 (country-of-origin: China vs France) factorial experiment. The participants (n=393) are recruited through the Amazon Mechanical Turk system (Mturks). Findings – The findings suggest that using Twitter substantially mitigates participants’ negative evaluation of the organization undergoing a crisis. Country-of-origin affects how individuals perceive the organization after it has experienced a crisis. In addition, participants’ product involvement intensifies the reputational threat specifically for the organization with a less favorable country-of-origin perception. Originality/value – This study is one of the few empirically based studies in international public relations research, using an experiment to extrapolate the effects of social media and country-of-origin on consumers’ crisis appraisal. This investigation reinforces the need to consider social media not just at the individual level, but also as a form of communication that can have broader consequences at the organizational level. In addition, it is important for company leaders to understand that the organization’s home country image may exacerbate the negative management outcomes during a crisis. It is expected that this study yields theoretically indicative, empirically informative, and culturally relevant results.


Sign in / Sign up

Export Citation Format

Share Document