AirCalypse: Can Twitter Help in Urban Air Quality Measurement and Who are the Influential Users?
Prithviraj Pramanik, Tamal Mondal, Subrata Nandi, Mousumi Saha
TL;DR
This work addresses urban air quality monitoring in Delhi by leveraging Twitter as a social sensor to augment sparse CPCB CAAQMS data. It introduces a TRank-based method to identify influential users through Retweet, Favorite, and Follower signals and analyzes how their pollution-related posts correlate with ground-truth $PM_{2.5}$ measurements. The findings indicate that focusing on influential users improves alignment between social perception and physical pollution data, suggesting a viable crowd-sensed augmentation of air-quality monitoring. The study also discusses practical challenges, notably geo-location limitations and normalization between user types, pointing toward an end-to-end, geo-aware social sensing framework for urban environments.
Abstract
In this digital age, Online Social Media's ubiquity has led it to it's role as a "Sensor". Starting from disaster response to political predictions, online social media like Twitter, have been instrumental and are actively researched areas. In this work, we have focused on something quite insidious in the current context, i.e., air pollution in developing regions. Starting as an empirical study on using Twitter as a "Sensor" to measure air quality, the focal point of this work is to identify the users who have been actively tweeting in the air pollution events in Delhi, the capital of India. From these users, we try to identify the influential ones, who play a significant role in creating the initial awareness and hence act as "Sensors". We have utilized a tailored "TRank" algorithm for finding out the influential users by considering \textit{Retweet, Favorite, and Follower influence} of the users. After ranking the users based on their social influence, we further study the behavior, i.e., perception of pollution from those users' posts with respect to the actual air pollution levels using the physical sensors. The tracking of influential users in air quality monitoring assists in developing a crowd sensed air quality measurement framework, which can augment the physical air quality sensors for raising awareness against air pollution.