Instademic: Deriving spatiotemporal contact networks from tagged social media images or We don’t care where you are, just who you were with

Identification of persons who have come in contact with COVID-19 and future viruses is necessary for early identification of potential carriers to improve public health. A Spatiotemporal Colocation Network (SCN) represents entities in terms of being present at the same location within a specified temporal window. A key factor of image sharing systems is user annotation, or tagging. By generating the network of collocated entities in image files across a temporal window we can map expected viral diffusion and identify potential carriers. Actual absolute location of entities is not important here. In colocation analysis for virus carrier detection “relative colocation” is important. i.e. we want to know that X was collocated with Y at time t, and Y was collocated with Z at time t’ which will enable us to build a “contact network”. Our proposed method uses tagging meta-data and does not require sharing of actual images.