On Localizing Urban Events with Instagram
On Localizing Urban Events with Instagram Prasanna Giridhar, Shiguang Wang, Tarek Abdelzaher Raghu Ganti Lance Kaplan, Jemin George Department of Computer Science IBM Research US Army Research Laboratory University of Illinois at Urbana Champaign Yorktown Heights, NY, USA 2800 Powder Mill Road Urbana, Illinois - 61801, USA Adelphi, MD 20783, USA Abstract—This paper develops an algorithm that exploits that allows anyone to search for Instagram images (using picture-oriented social networks to localize urban events. We tags or keywords). Hence, content access for a general event choose picture-oriented networks because taking a picture re- localization service becomes feasible. Second, unlike text- quires physical proximity, thereby revealing the location of the based social networks with publicly available content, such photographed event. Furthermore, most modern cell phones are as Twitter, Instagram features a content type that generally equipped with GPS, making picture location, and time metadata requires physical proximity to the event. While it is possible commonly available. We consider Instagram as the social network of choice and limit ourselves to urban events (noting that the to tweet about a volcano from across the globe, it is harder majority of the world population lives in cities). The paper to take pictures of it without physical proximity. Hence, the introduces a new adaptive localization algorithm that does not spatial distribution of Instagram content has a better correlation require the user to specify manually tunable parameters. We with actual event locations. Third, unlike other picture-based evaluate the performance of our algorithm for various real-world social networks, such as Flickr, Instagram content is much datasets, comparing it against a few baseline methods.
[Show full text]