Search Results (28)

Over the past decade, the frequency and impact of both natural and human-induced disasters have increased significantly, highlighting the urgent need for effective and timely relief operations. Disaster response requires efficient allocation of resources to the right locations and disaster types in a cost- and time-effective manner. However, during such events, large volumes of unverified and rapidly spreading information—especially on social media—often complicate situational awareness and decision-making. Consequently, extracting actionable insights and accurately classifying disaster-related information from social media platforms has become a critical research challenge. Machine Learning (ML) approaches have shown strong potential for categorizing disaster-related tweets, yet substantial variations in model accuracy persist across disaster types and regional contexts, suggesting that universal models may overlook linguistic and cultural nuances. This paper investigates the categorization and sub-categorization of natural disaster tweets using a labeled dataset of over 32,000 samples. Logistic Regression and Random Forest classifiers were trained and evaluated after comprehensive preprocessing to predict disaster categories and sub-categories. Furthermore, a country-specific prediction framework was implemented to assess how regional and cultural variations influence model performance. The results demonstrate strong overall classification accuracy, while revealing marked differences across countries, emphasizing the importance of context-aware, culturally adaptive ML approaches for reliable disaster information management. Full article

Natural disasters like wildfires pose significant threats to communities, which necessitates timely and effective disaster response strategies. While Aspect-based Sentiment Analysis (ABSA) has been widely used to extract sentiment-related information at the sub-sentence level, the corresponding field of Aspect-based Emotion Analysis (ABEA) remains underexplored due to dataset limitations and the increased complexity of emotion classification. In this study, we used EmoGRACE, a fine-tuned BERT-based model for ABEA, which we applied to georeferenced tweets of the 2020 California wildfires. The results for this case study reveal distinct spatio-temporal emotion patterns for wildfire-related aspect terms, with fear and sadness increasing near wildfire perimeters. This study demonstrates the feasibility of tracking emotion dynamics across disaster-affected regions and highlights the potential of ABEA in real-time disaster monitoring. The results suggest that ABEA can provide a nuanced understanding of public sentiment during crises for policymakers. Full article

In this research, we explore the use of Large Language Models (LLMs) and clustering techniques to automate the structuring and labeling of disaster-related social media content. With a gathered dataset comprising millions of tweets related to various disasters, our approach aims to transform unstructured and unlabeled data into a structured and labeled format that can be readily used for training machine learning algorithms and enhancing disaster response efforts. We leverage LLMs to preprocess and understand the semantic content of the tweets, applying several semantic properties to the data. Subsequently, we apply clustering techniques to identify emerging themes and patterns that may not be captured by predefined categories, with these patterns surfaced through topic extraction of the clusters. We proceed with manual labeling and evaluation of 10,000 examples to evaluate the LLMs’ ability to understand tweet features. Our methodology is applied to real-world data for disaster events, with results directly applicable to actual crisis situations. Full article

Earthquakes are sudden-onset disasters requiring rapid, accurate information for effective crisis response. Social media (SM) platforms provide abundant geospatial data but are often unstructured and produced by diverse users, posing challenges in filtering relevant content. Traditional content filtering methods rely on natural language processing (NLP), which underperforms with mixed-language posts or less widely spoken languages. Moreover, these approaches often neglect the spatial proximity of users to the event, a crucial factor in determining relevance during disasters. This study proposes an NLP-free model that assesses the spatial credibility of SM content by analysing users’ spatial trajectories. Using earthquake-related tweets, we developed a machine learning-based classification model that categorises posts as directly relevant, indirectly relevant, or irrelevant. The Random Forest model achieved the highest overall classification accuracy of 89%, while the k-NN model performed best for detecting directly relevant content, with an accuracy of 63%. Although promising overall, the classification accuracy for the directly relevant category indicates room for improvement. Our findings highlight the value of spatial analysis in enhancing the reliability of SM data (SMD) during crisis events. By bypassing textual analysis, this framework supports relevance classification based solely on geospatial behaviour, offering a novel method for evaluating content trustworthiness. This spatial approach can complement existing crisis informatics tools and be extended to other disaster types and event-based applications. Full article

The 2010 Haiti earthquake stands as one of the most catastrophic events in terms of loss of life and destruction. Following an earthquake, there is an urgent demand for information. Regrettably, few studies have tracked the progress of the post-disaster recovery, leaving this phase poorly understood. In previous years, data were exclusively collected through on-site missions, but today, social media (SM) has enhanced earthquake reconnaissance teams’ capacity to collect data beyond the emergency phase. However, text data from SM is unstructured, making it necessary to use natural language processing techniques to extract meaningful information. Sentiment analysis (SA), which classifies people’s opinions into positive, negative, or neutral polarity, is a promising tool for understanding earthquake recovery. For the purposes of this paper, we conduct SA at the tweet level on data collected around the tenth anniversary of the earthquake using human expertise to fine-tune automatic classification methods. We conclude that the anniversary date is the best time to collect data. In our sample, 56.3% of the tweets in the sample were classified as negative, followed by positive (27.3%), neutral (8.2%), and unrelated (8.1%). In our study, we conclude that the assessment of the recovery progress based on data collected from Twitter is negative. The automatic method for SA with the highest accuracy is ‘btweet’. The assessment result must be validated by stakeholders. Full article

During natural disasters, social media platforms, such as X (formerly Twitter), become a valuable source of real-time information, with eyewitnesses and affected individuals posting messages about the produced damage and the victims. Although this information can be used to streamline the intervention process of local authorities and to achieve a better distribution of available resources, manually annotating these messages is often infeasible due to time and cost constraints. To address this challenge, we explore the use of semi-supervised learning, a technique that leverages both labeled and unlabeled data, to enhance neural models for disaster tweet classification. Specifically, we investigate state-of-the-art semi-supervised learning models and focus on co-training, a less-explored approach in recent years. Moreover, we propose a novel hybrid co-training architecture, Multihead Average Pseudo-Margin, which obtains state-of-the-art results on several classification tasks. Our approach extends the advantages of the voting mechanism from Multihead Co-Training by using the Average Pseudo-Margin (APM) score to improve the quality of the pseudo-labels and self-adaptive confidence thresholds for improving imbalanced classification. Our method achieves up to 7.98% accuracy improvement in low-data scenarios and 2.84% improvement when using the entire labeled dataset, reaching 89.55% accuracy on the Humanitarian task and 91.23% on the Informative task. These results demonstrate the potential of our approach in addressing the critical need for automated disaster tweet classification. We made our code publicly available for future research. Full article

With the advancement of the Internet, social media platforms have gradually become powerful in spreading crisis-related content. Identifying informative tweets associated with natural disasters is beneficial for the rescue operation. When faced with massive text data, choosing the pivotal features, reducing the calculation expense, and increasing the model classification performance is a significant challenge. Therefore, this study proposes a multi-strategy improved black-winged kite algorithm (MSBKA) for feature selection of natural disaster tweets classification based on the wrapper method’s principle. Firstly, BKA is improved by utilizing the enhanced Circle mapping, integrating the hierarchical reverse learning, and introducing the Nelder–Mead method. Then, MSBKA is combined with the excellent classifier SVM (RBF kernel function) to construct a hybrid model. Finally, the MSBKA-SVM model performs feature selection and tweet classification tasks. The empirical analysis of the data from four natural disasters shows that the proposed model has achieved an accuracy of 0.8822. Compared with GA, PSO, SSA, and BKA, the accuracy is increased by 4.34%, 2.13%, 2.94%, and 6.35%, respectively. This research proves that the MSBKA-SVM model can play a supporting role in reducing disaster risk. Full article

The Internet’s development has prompted social media to become an essential channel for disseminating disaster-related information. Increasing the accuracy of emotional polarity recognition in tweets is conducive to the government or rescue organizations understanding the public’s demands and responding appropriately. Existing sentiment analysis models have some limitations of applicability. Therefore, this research proposes an IDBO-CNN-BiLSTM model combining the swarm intelligence optimization algorithm and deep learning methods. First, the Dung Beetle Optimization (DBO) algorithm is improved by adopting the Latin hypercube sampling, integrating the Osprey Optimization Algorithm (OOA), and introducing an adaptive Gaussian–Cauchy mixture mutation disturbance. The improved DBO (IDBO) algorithm is then utilized to optimize the Convolutional Neural Network—Bidirectional Long Short-Term Memory (CNN-BiLSTM) model’s hyperparameters. Finally, the IDBO-CNN-BiLSTM model is constructed to classify the emotional tendencies of tweets associated with the Hurricane Harvey event. The empirical analysis indicates that the proposed model achieves an accuracy of 0.8033, outperforming other single and hybrid models. In contrast with the GWO, WOA, and DBO algorithms, the accuracy is enhanced by 2.89%, 2.82%, and 2.72%, respectively. This study proves that the IDBO-CNN-BiLSTM model can be applied to assist emergency decision-making in natural disasters. Full article

In the early stages of a disaster caused by a natural hazard (e.g., flood), the amount of available and useful information is low. To fill this informational gap, emergency responders are increasingly using data from geo-social media to gain insights from eyewitnesses to build a better understanding of the situation and design effective responses. However, filtering relevant content for this purpose poses a challenge. This work thus presents a comparison of different machine learning models (Naïve Bayes, Random Forest, Support Vector Machine, Convolutional Neural Networks, BERT) for semantic relevance classification of flood-related, German-language Tweets. For this, we relied on a four-category training data set created with the help of experts from human aid organisations. We identified fine-tuned BERT as the most suitable model, averaging a precision of 71% with most of the misclassifications occurring across similar classes. We thus demonstrate that our methodology helps in identifying relevant information for more efficient disaster management. Full article

Climate change has led to a variety of disasters that have caused damage to infrastructure and the economy with societal impacts to human living. Understanding people’s emotions and stressors during disaster times will enable preparation strategies for mitigating further consequences. In this paper, we mine emotions and stressors encountered by people and shared on Twitter during Hurricane Harvey in 2017 as a showcase. In this work, we acquired a dataset of tweets from Twitter on Hurricane Harvey from 20 August 2017 to 30 August 2017. The dataset consists of around 400,000 tweets and is available on Kaggle. Next, a BERT-based model is employed to predict emotions associated with tweets posted by users. Then, natural language processing (NLP) techniques are utilized on negative-emotion tweets to explore the trends and prevalence of the topics discussed during the disaster event. Using Latent Dirichlet Allocation (LDA) topic modeling, we identified themes, enabling us to manually extract stressors termed as climate-change-related stressors. Results show that 20 climate-change-related stressors were extracted and that emotions peaked during the deadliest phase of the disaster. This indicates that tracking emotions may be a useful approach for studying environmentally determined well-being outcomes in light of understanding climate change impacts. Full article

Social media platforms are valuable data sources in the study of public reactions to events such as natural disasters and epidemics. This research assesses for selected countries around the globe the time lag between daily reports of COVID-19 cases and GDELT (Global Database of Events, Language, and Tone) and Twitter (X) COVID-19 mentions between February 2020 and April 2021 using time series analysis. Results show that GDELT articles and tweets preceded COVID-19 infections in Australia, Brazil, France, Greece, India, Italy, the U.S., Canada, Germany, and the U.K., while for Poland and the Philippines, tweets preceded and GDELT articles lagged behind COVID-19 disease incidences, respectively. This shows that the application of social media and news data for surveillance and management of pandemics needs to be assessed on a case-by-case basis for different countries. It also points towards the applicability of time series data analysis for only a limited number of countries due to strict data requirements (e.g., stationarity). A deviation from generally observed lag patterns in a country, i.e., periods with low COVID-19 infections but unusually high numbers of COVID-19-related GDELT articles or tweets, signals an anomaly. We use the seasonal hybrid extreme Studentized deviate test to detect such anomalies. This is followed by text analysis of news headlines from NewsBank and Google on the date of these anomalies to determine the probable event causing an anomaly, which includes elections, holidays, and protests. Full article

Social media platforms such as Twitter are a vital source of information during major events, such as natural disasters. Studies attempting to automatically detect textual communications have mostly focused on machine learning and deep learning algorithms. Recent evidence shows improvement in disaster detection models with the use of contextual word embedding techniques (i.e., transformers) that take the context of a word into consideration, unlike the traditional context-free techniques; however, studies regarding this model are scant. To this end, this paper investigates a selection of ensemble learning models by merging transformers with deep neural network algorithms to assess their performance in detecting informative and non-informative disaster-related Twitter communications. A total of 7613 tweets were used to train and test the models. Results indicate that the ensemble models consistently yield good performance results, with F-score values ranging between 76% and 80%. Simpler transformer variants, such as ELECTRA and Talking-Heads Attention, yielded comparable and superior results compared to the computationally expensive BERT, with F-scores ranging from 80% to 84%, especially when merged with Bi-LSTM. Our findings show that the newer and simpler transformers can be used effectively, with less computational costs, in detecting disaster-related Twitter communications. Full article

Natural hazard-related disasters are disruptive events with significant impact on people, communities, buildings, infrastructure, animals, agriculture, and environmental assets. The exponentially increasing anthropogenic activities on the planet have aggregated the climate change and consequently increased the frequency and severity of these natural hazard-related disasters, and consequential damages in cities. The digital technological advancements, such as monitoring systems based on fusion of sensors and machine learning, in early detection, warning and disaster response systems are being implemented as part of the disaster management practice in many countries and presented useful results. Along with these promising technologies, crowdsourced social media disaster big data analytics has also started to be utilized. This study aims to form an understanding of how social media analytics can be utilized to assist government authorities in estimating the damages linked to natural hazard-related disaster impacts on urban centers in the age of climate change. To this end, this study analyzes crowdsourced disaster big data from Twitter users in the testbed case study of Australian states and territories. The methodological approach of this study employs the social media analytics method and conducts sentiment and content analyses of location-based Twitter messages (n = 131,673) from Australia. The study informs authorities on an innovative way to analyze the geographic distribution, occurrence frequency of various disasters and their damages based on the geo-tweets analysis. Full article

The recent surge of social media networks has provided a channel to gather and publish vital medical and health information. The focal role of these networks has become more prominent in periods of crisis, such as the recent pandemic of COVID-19. These social networks have been the leading platform for broadcasting health news updates, precaution instructions, and governmental procedures. They also provide an effective means for gathering public opinion and tracking breaking events and stories. To achieve location-based analysis for social media input, the location information of the users must be captured. Most of the time, this information is either missing or hidden. For some languages, such as Arabic, the users’ location can be predicted from their dialects. The Arabic language has many local dialects for most Arab countries. Natural Language Processing (NLP) techniques have provided several approaches for dialect identification. The recent advanced language models using contextual-based word representations in the continuous domain, such as BERT models, have provided significant improvement for many NLP applications. In this work, we present our efforts to use BERT-based models to improve the dialect identification of Arabic text. We show the results of the developed models to recognize the source of the Arabic country, or the Arabic region, from Twitter data. Our results show 3.4% absolute enhancement in dialect identification accuracy on the regional level over the state-of-the-art result. When we excluded the Modern Standard Arabic (MSA) set, which is formal Arabic language, we achieved 3% absolute gain in accuracy between the three major Arabic dialects over the state-of-the-art level. Finally, we applied the developed models on a recently collected resource for COVID-19 Arabic tweets to recognize the source country from the users’ tweets. We achieved a weighted average accuracy of 97.36%, which proposes a tool to be used by policymakers to support country-level disaster-related activities. Full article

Social media has been a vital channel for communicating and broadcasting disaster-related information. However, the global spatiotemporal patterns of social media users’ activities, interactions, and connections after a natural disaster remain unclear. Hence, we integrated geocoding, geovisualization, and complex network methods to illustrate and analyze the online social network’s spatiotemporal evolution. Taking the super typhoon Haiyan as a case, we constructed a retweeting network and mapped this network according to the tweets’ location information. The results show that (1) the distribution of in-degree and out-degree follow power-law and retweeting networks are scale-free. (2) A local catastrophe could attract significant global interest but with strong geographical heterogeneity. The super typhoon Haiyan especially attracted attention from the United States, Europe, and Australia, in which users are more active in posting and forwarding disaster-related tweets than other regions (except the Philippines). (3) The users’ interactions and connections are also significantly different between countries and regions. Connections and interactions between the Philippines and the United States, Europe, and Australia were much closer than in other regions. Therefore, the agencies and platforms should also pay attention to other countries and regions outside the disaster area to provide more valuable information for the local people. Full article