For Journal Club Dr Farida Vis presented research that uses Twitter to improve food safety. Here, Grantham Scholar Patience Muchada, who is supervised by Farida, sets out the possibilities and limitations of using social media as research data.
Journal Club is chance for Scholars to take a multidisciplinary approach to sustainability work. This week it was led by my supervisor, Dr Farida Vis, from the Information School at the University of Sheffield. And she chose 2 articles for us to discuss. Firstly, nEmesis: Which restaurants should you avoid today? And the second was: Deploying nEmesis: Preventing Foodborne Illness by Data Mining Social Media.
Both of these papers take an innovative approach to food safety: using social media.
Global food security is one of the core research themes of the Grantham Centre. Food security means ensuring that everyone gets enough nutritious and safe food for a healthy lifestyle. So food safety is key to achieving this goal. But the number of food borne illnesses is unfortunately increasing. Nearly 420,000 people die annually from food borne illnesses all over the world.
Both articles chosen by Dr Vis highlight the economic cost of health losses resulting from foodborne illness: $78 billion per year in the US alone. But they also present a relatively novel approach to reducing food infections, Twitter.
The authors present a platform that uses Twitter data to help identify restaurants where health and food safety inspections might be urgently needed. This is done by using a user’s mobile phone location at the time a tweet was posted. Using the location, the authors track visits to nearby restaurants. Then they can track any food borne illness or related symptoms reported in tweets to those visits.
With everyone in the room owning a social media account of some sort, it was clear that it is a rich source of data. And it could give useful insights into human behaviour, practices and experiences. This is just what many (social) scientists are trying to understand.
More important however, is how the data is collected, analysed, interpreted, or manipulated to reach whatever claims or conclusions the researcher then makes. As such, we had a discussion on the assumptions made and methods used in the study.
Scholars asked a range of questions. For example, is it ethical to track someone’s movements this way? And what kind of accuracy can we really get from using tweets this way? One Scholar asked: “Does being within 50m of a restaurant imply that one has actually eaten there? And if the same user were to tweet about an upset stomach the next day, is that necessarily linked to the supposedly consumed restaurant food?”
Additional concerns were raised about the representativeness of the sample size. As one person put it: “who really tweets about such experiences?” Others questioned how much can trust what people do say, in other words performativity. “What people do and what they say they do are not always the same.” With a single tweet, there is hardly a way of verifying.
However, the authors claim to have prevented a statistically significant number of further cases of food poisoning by identifying problematic restaurants with their technology. But then again there was the classic failure of Google Flu Trends. This was another famous technology that made bold predictive claims. Overall, we questioned the scope of the success.
Despite all the critique, there was still some acknowledgement of the novelty of the work to use Twitter to improve food safety. And the authors’ gave a clear description of their methodology. So, while most Grantham Scholars present may not use social media data in their projects, it was good to be aware of the possibility and to see an example of it in use.
If the authors succeed in improving their technology and reduce the spread of food borne illnesses, then that is one step towards global food safety and security.
Journal Club is meet up of Grantham Scholars to discuss publications from a multidisciplinary perspective. It is part of the Grantham Scholar training programme.
Edited by Claire Moran. Photo by Tracy Le Blanc from Pexels.