Image Interventions: Developing Methods for Analyzing Social Media Images Across Academia and Government

• In Event: New Theories and Methods for the Study of Social Media Images Within and Beyond Academia

Abstract

In November 2016 one in five images shared by UK social media users were of food. Academics from a range of different disciplines, including Media and Communication have shown a keen interest in social media food images, but these images are also of increased interest to government researchers aiming to better understand how to deal with food safety issues as well as identifying new trends through social media image sharing. To understand how government researchers can develop robust and ethical methods for addressing this new data source, this paper reports on an innovative collaboration between the Visual Social Media Lab and the Food Standards Agency in the UK. Through four mini case studies different new innovative methodological approaches and tools developed in academia as well as industry are tested in order to build on the emerging new social media research capabilities within the FSA. Two case studies are specifically within the FSA’s remit: 1) an examination of videos and accompanying discussions on YouTube highlighting the perceived heath benefits and risks of raw (unpasteurized) milk consumption; and 2) images collected from a range of different platforms that potentially highlighting the consumption of undercooked chicken (which may contain campylobacter, which can lead to serious illness). Two new trends highlight the rise of images of vegan lifestyles on the one hand, thus focusing on specific new food trends and other the other, focus on new trends in technical capabilities. Specifically by working with new computational methods, Computer Vision in particular, and assessing the ability for these techniques to quickly identify food images, as well as images of specific foods within images. Overall, this paper is also concerned with critically evaluating these different approaches together and seeks to map out in more detail the types of questions each is able to answer and how this might align with the needs and remit of a government agency.

Author

• Anne Burns, U of Sheffield