Towards Detecting the Crowd Involved in Social Events

Abstract

Knowing how people interact with urban environments is fundamental for a variety of fields, ranging from transportation to social science. Despite the fact that human mobility patterns have been a major topic of study in recent years, a challenge to understand large-scale human behavior when a certain event occurs remains due to a lack of either relevant data or suitable approaches. Psychological crowd refers to a group of people who are usually located at different places and show different behaviors, but who are very sensitively driven to take the same act (gather together) by a certain event, which has been theoretically studied by social psychologists since the 19th century. This study aims to propose a computational approach using a machine learning method to model psychological crowds, contributing to the better understanding of human activity patterns under events. Psychological features and mental unity of the crowd are computed to detect the involved individuals. A national event happening across the USA in April, 2015 is analyzed using geotagged tweets as a case study to test our approach. The result shows that 81% of individuals in the crowd can be successfully detected. Through investigating the geospatial pattern of the involved users, not only can the event related users be identified but also those unobserved users before the event can be uncovered. The proposed approach can effectively represent the psychological feature and measure the mental unity of the psychological crowd, which sheds light on the study of large-scale psychological crowd and provides an innovative way to understanding human behavior under events.

Analysis of crowds' movement using Twitter

“…Huang et al 16 addressed the psychological aspect of the crowds by modeling their features semantically and using a machine learning to detect the individuals who show mental unity, detection what the authors referred to as a psychological crowd. Huang et al 16 addressed the psychological aspect of the crowds by modeling their features semantically and using a machine learning to detect the individuals who show mental unity, detection what the authors referred to as a psychological crowd.…”

“…Finally, we report on two proposals in the field of event detection based on social media but with a different aim. Huang et al addressed the psychological aspect of the crowds by modeling their features semantically and using a machine learning to detect the individuals who show mental unity, detection what the authors referred to as a psychological crowd. Secondly, in the work of Xu et al, the authors addressed specifically the issue of emergency event by analysis data from the Chinese social media platform Weibo.…”

A spatial‐temporal‐semantic approach for detecting local events using geo‐social media data

“…Topic modeling was then implemented based on a Java library, Mallet (McCallum, ), to infer the content of outlier tweets by investigating the topic distributions over outlier tweets and word distributions over topics. The number of topics set for training the LDA model was estimated by calculating perplexity (Equation 6), where 80% of outlier tweets were used for training and the remainder were used for evaluation (Huang, Fan, & Zipf, ). The perplexity distribution over the number of topics is shown in Figure .…”

Research on event perception based on geo-tagged social media data

“…Gao et al (2018) used heatmap to visualize extracted events from social media data. Huang et al (2017) aggregated data based on administrative regions, which is convenient for decisionmaking.…”