Extraction of Pluvial Flood Relevant Volunteered Geographic Information (VGI) by Deep Learning from User Generated Texts and Photos

Feng, Yilin; Sester, Monika

doi:10.3390/ijgi7020039

Cited by 77 publications

(24 citation statements)

References 37 publications

(51 reference statements)

Supporting

Mentioning

Contrasting

Unclassified

Order By: Relevance

“…Feng and Sester [18] used CNN and other methods to classify pluvial flood relevant tweets. Both text and photos in the tweets were combined and classified as relevant or irrelevant.…”

Section: Flooding Photo Classificationmentioning

confidence: 99%

“…As a non-manual and efficient filtering method, deep learning is a promising approach to extracting flood-relevant posts from massive social media data [17]. For example, recent studies [18][19][20] analyzed both the text and image of a post to determine whether the post is flood relevant or not. More importantly, the deep learning method can process the massive social media data in real-time, providing timely information for first response of the local disaster management team.…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Prototyping a Social Media Flooding Photo Screening System Based on Deep Learning

Ning

Hodgson

et al. 2020

IJGI

View full text Add to dashboard Cite

This article aims to implement a prototype screening system to identify flooding-related photos from social media. These photos, associated with their geographic locations, can provide free, timely, and reliable visual information about flood events to the decision-makers. This screening system, designed for application to social media images, includes several key modules: tweet/image downloading, flooding photo detection, and a WebGIS application for human verification. In this study, a training dataset of 4800 flooding photos was built based on an iterative method using a convolutional neural network (CNN) developed and trained to detect flooding photos. The system was designed in a way that the CNN can be re-trained by a larger training dataset when more analyst-verified flooding photos are being added to the training set in an iterative manner. The total accuracy of flooding photo detection was 93% in a balanced test set, and the precision ranges from 46–63% in the highly imbalanced real-time tweets. The system is plug-in enabled, permitting flexible changes to the classification module. Therefore, the system architecture and key components may be utilized in other types of disaster events, such as wildfires, earthquakes for the damage/impact assessment.

show abstract

“…Feng and Sester [18] used CNN and other methods to classify pluvial flood relevant tweets. Both text and photos in the tweets were combined and classified as relevant or irrelevant.…”

Section: Flooding Photo Classificationmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

Prototyping a Social Media Flooding Photo Screening System Based on Deep Learning

Ning

Hodgson

et al. 2020

IJGI

View full text Add to dashboard Cite

show abstract

“…where is the current word and c is the size of the window. After learning with a large corpus of Weibo posts, the weights were the corresponding vector representation for each word to train RNN model (Feng et al, 2018).…”

Section: Feature Representationmentioning

confidence: 99%

Online Urban Waterlogging Monitoring Based on Recurrent Neural Network for Classification of Microblogging Text

Liu¹,

Hao²,

Zhang³

et al. 2020

Preprint

View full text Add to dashboard Cite

Abstract. With the global climate change and rapid urbanization, urban flood disaster spreads and becomes increasingly serious in China. The urban rainstorm and waterlogging have become an urgent challenge that needs to be real-time monitored and further predicted for the improvement of urbanization construction. In this paper, we trained a recurrent neural network (RNN) model to classify microblogging posts related to urban waterlogging, and establish an online monitoring system of urban waterlogging caused by flood disaster. We manually curated more than 4,400 waterlogging posts to train the RNN model so that it can precisely identify waterlogging-related posts of Sina Weibo to timely find out urban waterlogging. The RNN model has been thoroughly evaluated, and our experimental results showed that it achieved higher accuracy than traditional machine learning methods, such as SVM and GBDT. Furthermore, we build a nationwide map of urban waterlogging based on recent two-year microblogging data.

show abstract

“…A text-based binary classification is required in order to extract more event-related information from the original data. With respect to a specific event, it is usually difficult to obtain large labeled data for deep learning in a short period of time, so this paper considers three robust machine learning algorithms (Naive Bayes, the k-nearest-neighbors algorithms, and Support Vector Machine) as alternatives [35][36][37][38]. The Naive Bayes (NB) method is a probabilistic classifier based on Bayes' theorem and assuming conditional independence.…”

Section: Extraction Of Event-related Informationmentioning

confidence: 99%

Geo-Tagged Social Media Data-Based Analytical Approach for Perceiving Impacts of Social Events

Zhu

Lin

Jendryke

et al. 2018

IJGI

View full text Add to dashboard Cite

Studying the impact of social events is important for the sustainable development of society. Given the growing popularity of social media applications, social sensing networks with users acting as smart social sensors provide a unique channel for understanding social events. Current research on social events through geo-tagged social media is mainly focused on the extraction of information about when, where, and what happened, i.e., event detection. There is a trend towards the machine learning of more complex events from even larger input data. This research work will undoubtedly lead to a better understanding of big geo-data. In this study, however, we start from known or detected events, raising further questions on how they happened, how they affect people’s lives, and for how long. By combining machine learning, natural language processing, and visualization methods in a generic analytical framework, we attempt to interpret the impact of known social events from the dimensions of time, space, and semantics based on geo-tagged social media data. The whole analysis process consists of four parts: (1) preprocessing; (2) extraction of event-related information; (3) analysis of event impact; and (4) visualization. We conducted a case study on the “2014 Shanghai Stampede” event on the basis of Chinese Sina Weibo data. The results are visualized in various ways, thus ensuring the feasibility and effectiveness of our proposed framework. Both the methods and the case study can serve as decision references for situational awareness and city management.

show abstract

Extraction of Pluvial Flood Relevant Volunteered Geographic Information (VGI) by Deep Learning from User Generated Texts and Photos

Cited by 77 publications

References 37 publications

Prototyping a Social Media Flooding Photo Screening System Based on Deep Learning

Prototyping a Social Media Flooding Photo Screening System Based on Deep Learning

Online Urban Waterlogging Monitoring Based on Recurrent Neural Network for Classification of Microblogging Text

Geo-Tagged Social Media Data-Based Analytical Approach for Perceiving Impacts of Social Events

Contact Info

Product

Resources

About