Urban Crowdsensing using Social Media: An Empirical Study on Transformer and Recurrent Neural Networks

Abstract

An important aspect of urban planning is understanding crowd levels at various locations, which typically require the use of physical sensors. Such sensors are potentially costly and time consuming to implement on a large scale. To address this issue, we utilize publicly available social media datasets and use them as the basis for two urban sensing problems, namely event detection and crowd level prediction. One main contribution of this work is our collected dataset from Twitter and Flickr, alongside ground truth events. We demonstrate the usefulness of this dataset with two preliminary supervised learning approaches: firstly, a series of neural network models to determine if a social media post is related to an event and secondly a regression model using social media post counts to predict actual crowd levels. We discuss preliminary results from these tasks and highlight some challenges.

Urban Heat Islands: Beating the Heat with Multi-Modal Spatial Analysis

“…For example, one can develop machine learning models and systems to predict the heat levels in different parts of a country, based on multi-modal data from different sources [48]. In addition to sensor and map-based data, one can enhance such measurements and predictions with crowd-sourced data from social media [49], mobile phones [48] and other relevant human sensors.…”

Photozilla: A Large-Scale Photography Dataset and Visual Embedding for 20 Photography Styles

“…We intend to collect more information and fuse different types of features, such as captions, locations, and hashtags [6,13], and perform multi-modal analysis and sequence prediction tasks for various applications of the dataset, such as next-POI (Point-Of-Interest) recommendations [14].…”