Quantifying people's experience during flood events with implications for hazard risk communication.

Nataliya Tkachenko,Stephen Jarvis,Rob Procter,Jichang Zhao

doi:10.1371/journal.pone.0244801

Abstract

Semantic drift is a well-known concept in distributional semantics, which is used to demonstrate gradual, long-term changes in meanings and sentiments of words and is largely detectable by studying the composition of large corpora. In our previous work, which used ontological relationships between words and phrases, we established that certain kinds of semantic micro-changes can be found in social media emerging around natural hazard events, such as floods. Our previous results confirmed that semantic drift in social media can be used to for early detection of floods and to increase the volume of ‘useful’ geo-referenced data for event monitoring. In this work we use deep learning in order to determine whether images associated with ‘semantically drifted’ social media tags reflect changes in crowd navigation strategies during floods. Our results show that alternative tags can be used to differentiate naïve and experienced crowds witnessing flooding of various degrees of severity.

Highlights

Unusual events and changes in the natural environment can significantly impact people’s dayto-day activities, information on human mobility has been primarily valued for its crucial role in response to disaster and evacuation strategies [1]
Neutral lexemes, which have previously demonstrated a transient shift of meaning around flood events [23], show an increased structural dissimilarity with both sets of words and this distance gradually decreases with the increase of event severity, for both cases before and after official risk communication messages
This can be indicative of the fact that during the early stages of flood events, lexemes that are prone to semantic drift under the influence of an approaching hazard are associated with different sets of scenes and, as the hazard evolves, the similarity between scenes increases

Summary

Introduction

Unusual events and changes in the natural environment can significantly impact people’s dayto-day activities, information on human mobility has been primarily valued for its crucial role in response to disaster and evacuation strategies [1]. Some studies have reported that the success of planning and executing evacuation operations to a great extent depend on exact information of where people are [2, 3]; other studies mention that real-time designation of the risk areas could benefit from the human movement patterns [4]. Successful geotargeting of appropriate shelter locations relies on ‘hot-spots’, that is vulnerable gatherings of people [5, 6], whereas adaptation of early and real-time warning communication to mobile outdoor populations can be instrumental for the deployment of a new generation of smart alert systems [7, 8].

Methods

Results

Discussion

Conclusion