Can Instagram posts help characterize urban micro-events?

• Main Authors: JAYARAJAH, Kasthuri, MISRA, Archan
• Format: text
• Language: English
• Published: Institutional Knowledge at Singapore Management University 2016
• Subjects: Semantics; Twitter; Feature extraction; Spatiotemporal phenomena; Media; Sensors; Event detection; Information fusion; Social networking (online); Databases and Information Systems; Social Media
• Online Access: https://ink.library.smu.edu.sg/sis_research/3639; https://ink.library.smu.edu.sg/context/sis_research/article/4641/viewcontent/fusion16_kjayarajah.pdf
• Institution: Singapore Management University

Social media content, from platforms such as Twitter and Foursquare, has enabled an exciting new field of social sensing, where participatory content generated by users has been used to identify unexpected emerging or trending events. In contrast to such text-based channels, we focus on image-sharing social applications (specifically Instagram), and investigate how such urban social sensing can leverage upon the additional multi-modal, multimedia content. Given the significantly higher fraction of geotagged content on Instagram, we aim to use such channels to go beyond identification of long-lived events (e.g., a marathon) to achieve finer-grained characterization of multiple micro-events (e.g., a person winning the marathon) that occur over the lifetime of the macro-event. Via empirical analysis from a corpus of Instagram data from 3 international marathons, we establish the need for novel data pre-processing as: (a) semantic annotation of image content indeed provides additional features distinct from text captions, and (b) an appreciable fraction of the posted images do not pertain to the event under consideration. We propose a framework, called EiM, that combines such preprocessing with clustering-based event detection. We show that our initial prototype of EiM shows promising results: it is able to identify many micro-events in the three marathons, with spatial and temporal resolution that is less than 1% and 10%, respectively, of the corresponding ranges for the macro-event.