Geo-tagged Tweets Research Articles

An exploratory tag map for attributes-in-space tasks

Geo-text data, which combine geographical locations with textual information (e.g., geo-tagged tweets), are typically visualized using tag maps. Since tags are rich in attribute information, tag maps are an intuitive method of visualizing how attribute domains carried by tags vary across space. However, users may be interested not only in the overall spatial distribution of tags but also in exploring detailed attributes-in-space analyses, such as examining how a subclass of attribute domains is distributed globally or checking whether all attribute subclasses exhibit the same global distribution pattern. To date, the methods for representing tags with visual encoding (e.g., size, color) to extend various attributes-in-space tasks to support exploratory analysis remain unclear. In this work, we extended tag maps to support exploratory analysis by distinguishing space searching into local or global spaces and attribute domains into within or between attribute classes, supporting four types of attributes-in-space tasks: global-within, local-within, global-between, and local-between tasks. We evaluated our exploratory tag map through two case studies: investigating major disaster occurrences from 1981 to 2020 and examining the leading causes of death in 2000 and 2019 for Spain, France, Germany and Italy. We used eye-tracking and a questionnaire to evaluate our exploratory tag map for comparison. Both methods had similar self-reported usability scores in terms of aesthetics, density, layout, and legibility. However, our exploratory tag map was more effective and efficient and had a lower cognitive load.

Understanding the impact of geotagging on location inference models for accurate generalization to non-geotagged datasets

For a location inference model to be successful, the properties of the geotagged tweets where the location inference model is developed and those of the non-geotagged tweets where the location inference model is applied need to match. We investigated location mentions within the tweet text field of 3,953,166 geotagged and 2,783,609 non-geotagged tweets across five of the most prominent Twitter sources. Specifically, we compared the frequency and the location entity types used to infer the locations within the two datasets. Overall we found statistically significant differences in location mentions between the two datasets. However, although statistically significant, thirteen of the fifteen analysed location entities, showed low effect sizes. We conclude that location inference models trained on geotagged datasets can generalize a non-geotagged dataset if special adjustments are made on the development of the location inference models.

Quantifying generalized trust in individuals and counties using language

Trust is predictive of civic cooperation and economic growth. Recently, the U.S. public has demonstrated increased partisan division and a surveyed decline in trust in institutions. There is a need to quantify individual and community levels of trust unobtrusively and at scale. Using observations of language across more than 16,000 Facebook users, along with their self-reported generalized trust score, we develop and evaluate a language-based assessment of generalized trust. We then apply the assessment to more than 1.6 billion geotagged tweets collected between 2009 and 2015 and derive estimates of trust across 2,041 U.S. counties. We find generalized trust was associated with more affiliative words (love, we, and friends) and less angry words (hate and stupid) but only had a weak association with social words primarily driven by strong negative associations with general othering terms (“they” and “people”). At the county level, associations with the Centers for Disease Control and Prevention (CDC) and Gallup surveys suggest that people in high-trust counties were physically healthier and more satisfied with their community and their lives. Our study demonstrates that generalized trust levels can be estimated from language as a low-cost, unobtrusive method to monitor variations in trust in large populations.

GeoCovaxTweets: A global analysis of COVID-19 vaccines and vaccination discourse on social media

Social media platforms play a crucial role in providing valuable information during crises, such as pandemics. The COVID-19 pandemic has created a global public health crisis, and vaccines are the key preventive measure for achieving herd immunity. However, some individuals use social media to oppose vaccines, undermining government efforts to eliminate the virus. This study introduces the “GeoCovaxTweets” dataset, consisting of 1.8 million geotagged tweets related to COVID-19 vaccines from January 2020 to November 2022, originating from 233 countries and territories. Each tweet includes state and country information, enabling researchers to analyze global spatial and temporal patterns. An extensive set of analyses are performed on the dataset to identify prominent topic clusters and explore public opinions across different vaccines and vaccination contexts. The study outlines the dataset curation methodology and provides instructions for local reproduction. We anticipate that the dataset will be valuable for crisis computing researchers, facilitating the exploration of Twitter conversations surrounding COVID-19 vaccines and vaccination, including trends, opinion shifts, misinformation, and anti-vaccination campaigns.

From an election to an insurrection: Investigating differential engagement and sentiment in the #defundthepolice and #defendthepolice network on Twitter.

Social movements and their respective countermovements have evolved to use online social media platforms to recruit followers, share pertinent information, discuss relevant issues, and draw the attention of political figures. Movements' strategic use of Twitter has increasingly been studied, though there are relatively few studies that compare social movements and their corresponding countermovements simultaneously. We examine engagement in the #DefundthePolice social movement and #DefendthePolice countermovement in a Twitter network comprised of retweets using both hashtags from August 2020 to January 2021. Text and sentiment analysis as well as a content analysis of a random sample of retweets in the network's 20 largest subgroups reveal four key patterns. First, information commonly communicated in historical social movements is communicated in the online, Twitter network. Second, the use of movement and countermovement hashtags to criticize is common, suggesting Twitter engagement with the movement/countermovement is not a sufficient indicator of support for the movement. Third, social movements are inextricably embedded in politics, with political discourse present in all the 20 largest subgroups. Finally, though we do not include geo-tagged tweets in the analysis, physical geography is key theme in multiple subgroups. Broadly, our findings demonstrate the breadth of topics communicated within movement networks and highlight the importance of qualitatively examining Twitter data in the study of social movements.

The Russian war in Ukraine increased Ukrainian language use on social media

The use of language is innately political, often a vehicle of cultural identity and the basis for nation building. Here, we examine language choice and tweeting activity of Ukrainian citizens based on 4,453,341 geo-tagged tweets from 62,712 users before and during the Russian war in Ukraine, from January 2020 to October 2022. Using statistical models, we disentangle sample effects, arising from the in- and outflux of users on Twitter (now X), from behavioural effects, arising from behavioural changes of the users. We observe a steady shift from the Russian language towards Ukrainian already before the war, which drastically speeds up with its outbreak. We attribute these shifts in large part to users’ behavioural changes. Notably, our analysis shows that more than half of the Russian-tweeting users switch towards Ukrainian with the Russian invasion. We interpret these findings as users’ conscious choice towards a more Ukrainian (online) identity and self-definition of being Ukrainian.

Characterizing environmental pollution with civil complaints and social media data: A case of the Greater Taipei Area

Environmental pollution is a major cause of nuisance and ill health among urban residents. Complaints are traditionally self-reported through phone-based systems. Social media provide novel channels to detect pollution-related incidents; however, their reliability has not been sufficiently evaluated. This study aimed to compare pollution incidents expressed on Twitter with those extracted from phone-based systems and to identify the built environment and socioeconomic attributes that can predict the likelihood of pollution incidents. A total of 639,746 tweets were retrieved from the Greater Taipei Area in 2017 and 110,716 self-reported pollution incidents were extracted from the Public Nuisance Petition system during the same period. The results suggest that complaints collected from phone-based systems and Twitter were found to have correlated with each other spatially, albeit they differ in temporal profiles and by the proportion of pollution categories. Catering businesses and the entertainment activities they attract appear to be the main sources of pollution complaints and can be precisely captured by geotagged tweets. This can serve as a strong predictor for pollution incidents, more than traditional indicators such as population density or industrial activities, as suggested by earlier studies. Social media analytics, with their ability to monitor and analyze online discussions in a timely manner, can be a valuable supplement to existing phone-based pollution monitoring procedures. The methodologies developed in this study have the potential to support the proactive management of urban environmental pollution, in which resources can be prioritized in key areas to further enhance the quality of urban services.

Twitter Sentiment Geographical Index Dataset

Promoting well-being is one of the key targets of the Sustainable Development Goals at the United Nations. Many national and city governments worldwide are incorporating Subjective Well-Being (SWB) indicators into their agenda, to complement traditional objective development and economic metrics. In this study, we introduce the Twitter Sentiment Geographical Index (TSGI), a location-specific expressed sentiment database with SWB implications, derived through deep-learning-based natural language processing techniques applied to 4.3 billion geotagged tweets worldwide since 2019. Our open-source TSGI database represents the most extensive Twitter sentiment resource to date, encompassing multilingual sentiment measurements across 164 countries at the admin-2 (county/city) level and daily frequency. Based on the TSGI database, we have created a web platform allowing researchers to access the sentiment indices of selected regions in the given time period.

Assessing place experiences in Luton and Darlington on Twitter with topic modelling and AI-generated lexicons

PurposeThe purpose of this paper is to examine and compare the in situ place experiences of people in Luton and Darlington.Design/methodology/approachThe study used 109,998 geotagged tweets from Luton and Darlington between 2020 and 2022 and conducted topic modelling using latent Dirichlet allocation. Lexicons were created using GPT-4 to evaluate the eight dimensions of place experience for each topic.FindingsThe study found that Darlington had higher counts in the sensorial, behavioural, designed and mundane dimensions of place experience than Luton. Conversely, Luton had a higher prevalence of the affective and intellectual dimensions, attributed to political and faith-related tweets.Originality/valueThe study introduces a novel approach that uses AI-generated lexicons for place experience. These lexicons cover four facets, two intentions and two intensities of place experience, enabling detection of words from any domain. This approach can be useful not only for town and destination brand managers but also for researchers in any field.

Staying Home, Tweeting Hope: Mixed Methods Study of Twitter Sentiment Geographical Index During US Stay-At-Home Orders.

Stay-at-home orders were one of the controversial interventions to curb the spread of COVID-19 in the United States. The stay-at-home orders, implemented in 51 states and territories between March 7 and June 30, 2020, impacted the lives of individuals and communities and accelerated the heavy usage of web-based social networking sites. Twitter sentiment analysis can provide valuable insight into public health emergency response measures and allow for better formulation and timing of future public health measures to be released in response to future public health emergencies. This study evaluated how stay-at-home orders affect Twitter sentiment in the United States. Furthermore, this study aimed to understand the feedback on stay-at-home orders from groups with different circumstances and backgrounds. In addition, we particularly focused on vulnerable groups, including older people groups with underlying medical conditions, small and medium enterprises, and low-income groups. We constructed a multiperiod difference-in-differences regression model based on the Twitter sentiment geographical index quantified from 7.4 billion geo-tagged tweets data to analyze the dynamics of sentiment feedback on stay-at-home orders across the United States. In addition, we used moderated effects analysis to assess differential feedback from vulnerable groups. We combed through the implementation of stay-at-home orders, Twitter sentiment geographical index, and the number of confirmed cases and deaths in 51 US states and territories. We identified trend changes in public sentiment before and after the stay-at-home orders. Regression results showed that stay-at-home orders generated a positive response, contributing to a recovery in Twitter sentiment. However, vulnerable groups faced greater shocks and hardships during the COVID-19 pandemic. In addition, economic and demographic characteristics had a significant moderating effect. This study showed a clear positive shift in public opinion about COVID-19, with this positive impact occurring primarily after stay-at-home orders. However, this positive sentiment is time-limited, with 14 days later allowing people to be more influenced by the status quo and trends, so feedback on the stay-at-home orders is no longer positively significant. In particular, negative sentiment is more likely to be generated in states with a large proportion of vulnerable groups, and the policy plays a limited role. The pandemic hit older people, those with underlying diseases, and small and medium enterprises directly but hurt states with cross-cutting economic situations and more complex demographics over time. Based on large-scale Twitter data, this sociological perspective allows us to monitor the evolution of public opinion more directly, assess the impact of social events on public opinion, and understand the heterogeneity in the face of pandemic shocks.

Climatic and Economic Background Determine the Disparities in Urbanites' Expressed Happiness during the Summer Heat.

Climate-change-induced extreme weather events increase heat-related mortality and health risks for urbanites, which may also affect urbanites' expressed happiness (EH) and well-being. However, the links among EH, climate, and socioeconomic factors remain unclear. Here we collected ∼6 million geotagged tweets from 44 Chinese prefecture-level cities based on Sina Weibo and performed a quadratic regression model to explore the relationships between summer heat and EH. A three-stage analysis was developed to examine spatiotemporal heterogeneity and identify factors contributing to disparities in urbanites' EH. Results show that all cities exhibited a similar hump-shaped relationship, with an overall optimal temperature (OT) of 22.8 °C. The estimated OT varied geographically, with 25.3, 23.8, and 20.0 °C from north to south. Moreover, a 1 standard deviation increase in heatwave intensity was associated with a 0.813 (95% CI: 0.177, 1.449) standard deviation decrease in EH. Notably, within the geographic scope of this study, it was observed that urbanites in northern China and economically underdeveloped cities faced significantly lower heat risks during the summer heat. This research provides insight for future studies and practical applications concerning extreme weather events, urbanites' mental health, and sustainable urban development goal.

A Modified Boosted Ensemble Classifier on Location Based Social Networking

One of the research issues that researchers are interested in is unbalanced dataclassification techniques. Boosting approaches like Wang's Boosting and ModifiedBoosted SVM (MBSVM) have been demonstrated to be more effective for unbalanceddata. Our proposal The Modified Boosted Random Forest (MBRF) classifier is aRandom Forest classifier that uses the Boosting approach. The main motivation of thestudy is to analyze sentiment of geotagged tweets understanding the state of mind ofpeople at FIFA and Olympics datasets. Tree based model Random Forest algorithmusing boosting approach classifies the tweets to build a recommendation system withan idea of providing commercial suggestions to participants, recommending localplaces to visit or perform activities. MBRF employs various strategies: i) a distancebasedweight-update method based on K-Medoids ii) a sign-based classifier eliminationtechnique. We have equally partitioned the datasets as 70% of data allocated fortraining and the remaining 30% data as test data. Our imbalanced data ratio measured3.1666 and 4.6 for FIFA and Olympics datasets. We looked at accuracy, precision,recall and ROC curves for each event. The average AUC achieved by MBRF on FIFAdataset is 0.96 and Olympics is 0.97. A comparison of MBRF and Decision tree modelusing 'Entropy' proved MBRF better.

A deep learning-based sentiment analysis approach (MF-CNN-BILSTM) and topic modeling of tweets related to the Ukraine–Russia conflict

Twitter, one of the most significant social media platforms, can be used as data sources to research public opinion on various topics, including political conflicts. People worldwide have expressed their opinions about the war between Russia and Ukraine since it began. The motivation of this study is to use deep learning techniques to reveal qualitative and quantitatively narrow-scoped situational awareness and emotional tendencies during crisis periods. In order to achieve this, a sizable dataset of geotagged tweets was gathered with specific terms associated with the conflict between Ukraine and Russia. Then, deep learning-based text mining techniques like topic modeling and sentiment analysis were applied to investigate people’s perspectives on conflict and emotional tendencies. The study will establish an impartial data source for the reports and articles of unbiased press members. This study used Valence Aware Dictionary and sEntiment Reasoner (VADER) to categorize the emotions related to the conflict between Russia and Ukraine in tweets. In addition, Latent Dirichlet Allocation (LDA) was used to extract various discussion topics. These techniques reveal the role of Twitter and compare and analyze the emotions and attitudes expressed on Twitter by different countries during the Ukraine–Russia conflict. In addition, a new deep learning-based sentiment classification “MF-CNN-BiLSTM” model that predicts and analyzes sentiments was proposed in this study. The proposed model stands for Multistage Feature Extraction using Convolutional Neural Network (CNN) and Bidirectional Long Short-Term Memory (BiLSTM). By combining various qualities and benefits of CNNs and BiLSTM, the suggested model makes it possible to identify both short- and long-term dependencies in ordinal data. The proposed model includes MF steps that strengthen feature extraction by identifying local dependencies. Experimental results demonstrate that it is possible to obtain more accurate sentiment classification results by the proposed method.

Detecting informal green, blue, and street physical activity spaces in the city using geotagged sports-related Twitter tweets.

Finland's natural physical environment and climate support a wide variety of informal outdoor sports, thereby motivating the population to do physical exercise in scenic environments. The vast majority of Finns enjoys outdoor recreational activities, and could thus be encouraged to post accounts of their year-round activities on social media. Our aim was to find out in what kind of areas and spaces, spatially, users are tweeting about sporting activities. We use geotagged Twitter tweets filtering for 16 sporting activity keywords in both English and Finnish. The case study was conducted in the Helsinki Metropolitan Area, Finland, with an emphasis on cross-country skiing as a sports activity when there is snow. In a secondary analysis we concentrated on the sports people were practicing in these locations when there was no snow. The location spaces are split in to three land cover types: green, blue, and street spaces. We found that approximately half of the 150 skiing-related tweets were geotagged in green spaces, and half in street spaces. This finding related to street space was attributable to a spatial scale error: when we checked the results manually we noticed that they referenced the sporting location in the green space. Hence, then over 90% of the 745 non-ski-related tweets were geotagged in a street space. We conclude that Twitter is a beneficial tool for detecting spaces used for informal physical activity. A shortcoming in current Finnish national sporting policies is that spaces for informal physical activity are not explicitly mentioned- we use the term informal with reference both to the space and to the sporting activity, whereby public spaces are used for physical activity. This new knowledge of sporting locations will help city planners and sports planners to improve informal sports facilities, which in turn will promote healthy exercise in cities.

Mpox Panic, Infodemic, and Stigmatization of the Two-Spirit, Lesbian, Gay, Bisexual, Transgender, Queer or Questioning, Intersex, Asexual Community: Geospatial Analysis, Topic Modeling, and Sentiment Analysis of a Large, Multilingual Social Media Database

The global Mpox (formerly, Monkeypox) outbreak is disproportionately affecting the gay and bisexual men having sex with men community. The aim of this study is to use social media to study country-level variations in topics and sentiments toward Mpox and Two-Spirit, Lesbian, Gay, Bisexual, Transgender, Queer or Questioning, Intersex, Asexual (2SLGBTQIAP+)-related topics. Previous infectious outbreaks have shown that stigma intensifies an outbreak. This work helps health officials control fear and stop discrimination. In total, 125,424 Twitter and Facebook posts related to Mpox and the 2SLGBTQIAP+ community were extracted from May 1 to December 25, 2022, using Twitter application programming interface academic accounts and Facebook-scraper tools. The tweets' main topics were discovered using Latent Dirichlet Allocation in the sklearn library. The pysentimiento package was used to find the sentiments of English and Spanish posts, and the CamemBERT package was used to recognize the sentiments of French posts. The tweets' and Facebook posts' languages were understood using the Twitter application programming interface platform and pycld3 library, respectively. Using ArcGis Online, the hot spots of the geotagged tweets were identified. Mann-Whitney U, ANOVA, and Dunn tests were used to compare the sentiment polarity of different topics and countries. The number of Mpox posts and the number of posts with Mpox and 2SLGBTQIAP+ keywords were 85% correlated (P<.001). Interestingly, the number of posts with Mpox and 2SLGBTQIAP+ keywords had a higher correlation with the number of Mpox cases (correlation=0.36, P<.001) than the number of posts on Mpox (correlation=0.24, P<.001). Of the 10 topics, 8 were aimed at stigmatizing the 2SLGBTQIAP+ community, 3 of which had a significantly lower sentiment score than other topics (ANOVA P<.001). The Mann-Whitney U test shows that negative sentiments have a lower intensity than neutral and positive sentiments (P<.001) and neutral sentiments have a lower intensity than positive sentiments (P<.001). In addition, English sentiments have a higher negative and lower neutral and positive intensities than Spanish and French sentiments (P<.001), and Spanish sentiments have a higher negative and lower positive intensities than French sentiments (P<.001). The hot spots of the tweets with Mpox and 2SLGBTQIAP+ keywords were recognized as the United States, the United Kingdom, Canada, Spain, Portugal, India, Ireland, and Italy. Canada was identified as having more tweets with negative polarity and a lower sentiment score (P<.04). The 2SLGBTQIAP+ community is being widely stigmatized for spreading the Mpox virus on social media. This turns the community into a highly vulnerable population, widens the disparities, increases discrimination, and accelerates the spread of the virus. By identifying the hot spots and key topics of the related tweets, this work helps decision makers and health officials inform more targeted policies.

Twitter mobility dynamics during the COVID-19 pandemic: A case study of London.

The current COVID-19 pandemic has profoundly impacted people's lifestyles and travel behaviours, which may persist post-pandemic. An effective monitoring tool that allows us to track the level of change is vital for controlling viral transmission, predicting travel and activity demand and, in the long term, for economic recovery. In this paper, we propose a set of Twitter mobility indices to explore and visualise changes in people's travel and activity patterns, demonstrated through a case study of London. We collected over 2.3 million geotagged tweets in the Great London Area (GLA) from Jan 2019 -Feb 2021. From these, we extracted daily trips, origin-destination matrices, and spatial networks. Mobility indices were computed based on these, with the year 2019 as a pre-Covid baseline. We found that in London, (1) People are making fewer but longer trips since March 2020. (2) In 2020, travellers showed comparatively reduced interest in central and sub-central activity locations compared to those in outer areas, whereas, in 2021, there is a sign of a return to the old norm. (3) Contrary to some relevant literature on mobility and virus transmission, we found a poor spatial relationship at the Middle Layer Super Output Area (MSOA) level between reported COVID-19 cases and Twitter mobility. It indicated that daily trips detected from geotweets and their most likely associated social, exercise and commercial activities are not critical causes for disease transmission in London. Aware of the data limitations, we also discuss the representativeness of Twitter mobility by comparing our proposed measures to more established mobility indices. Overall, we conclude that mobility patterns obtained from geo-tweets are valuable for continuously monitoring urban changes at a fine spatiotemporal scale.

American cultural regions mapped through the lexical analysis of social media

Cultural areas represent a useful concept that cross-fertilizes diverse fields in social sciences. Knowledge of how humans organize and relate their ideas and behavior within a society can help us to understand our actions and attitudes toward different issues. However, the selection of common traits that shape a cultural area is somewhat arbitrary. What is needed is a method that can leverage the massive amounts of data coming online, especially through social media, to identify cultural regions without ad-hoc assumptions, biases, or prejudices. This work takes a crucial step in this direction by introducing a method to infer cultural regions based on the automatic analysis of large datasets from microblogging posts. The approach presented here is based on the principle that cultural affiliation can be inferred from the topics that people discuss among themselves. Specifically, regional variations in written discourse are measured in American social media. From the frequency distributions of content words in geotagged tweets, the regional hotspots of words’ usage are found, and from there, principal components of regional variation are derived. Through a hierarchical clustering of the data in this lower-dimensional space, this method yields clear cultural areas and the topics of discussion that define them. It uncovers a manifest North–South separation, which is primarily influenced by the African American culture, and further contiguous (East–West) and non-contiguous divisions that provide a comprehensive picture of modern American cultural areas.

Enhanced geocoding precision for location inference of tweet text using spaCy, Nominatim and Google Maps. A comparative analysis of the influence of data selection.

Twitter location inference methods are developed with the purpose of increasing the percentage of geotagged tweets by inferring locations on a non-geotagged dataset. For validation of proposed approaches, these location inference methods are developed on a fully geotagged dataset on which the attached Global Navigation Satellite System coordinates are used as ground truth data. Whilst a substantial number of location inference methods have been developed to date, questions arise pertaining the generalizability of the developed location inference models on a non-geotagged dataset. This paper proposes a high precision location inference method for inferring tweets' point of origin based on location mentions within the tweet text. We investigate the influence of data selection by comparing the model performance on two datasets. For the first dataset, we use a proportionate sample of tweet sources of a geotagged dataset. For the second dataset, we use a modelled distribution of tweet sources following a non-geotagged dataset. Our results showed that the distribution of tweet sources influences the performance of location inference models. Using the first dataset we outweighed state-of-the-art location extraction models by inferring 61.9%, 86.1% and 92.1% of the extracted locations within 1 km, 10 km and 50 km radius values, respectively. However, using the second dataset our precision values dropped to 45.3%, 73.1% and 81.0% for the same radius values.

Impact of COVID-19 on online grocery shopping discussion and behavior reflected from Google Trends and geotagged tweets

People express opinions, make connections, and disseminate information on social media platforms. We considered grocery-related tweets as a proxy for grocery shopping behaviors or intentions. We collected data from January 2019 to January 2022, representing three typical times of the normal period before the COVID-19 pandemic, the outbreak period, and the widespread period. We obtained grocery-related geotagged tweets using a search term index based on the top 10 grocery chains in the US and compiled Google Trends online grocery shopping data. We performed a topic modeling analysis using the Latent Dirichlet Allocation (LDA), and verified that most of the collected tweets were related to grocery-shopping demands or experiences. Temporal and geographical analyses were applied to investigate when and where people talked more about groceries, and how COVID-19 affected them. The results show that the pandemic has been gradually changing people’s daily shopping concerns and behaviors, which have become more spread throughout the week since the pandemic began. Under the causal impact of COVID-19, people first experienced panic buying groceries followed by pandemic fatigue a year later. The normalized tweet counts show a decrease of 40% since the pandemic began, and the negative causal effect can be considered statistically significant (p-value = 0.001). The variation in the quantity of grocery-related tweets also reflects geographic diversity in grocery concerns. We found that people in non-farm areas with less population and relatively lower levels of educational attainment tend to act more sensitively to the evolution of the pandemic. Utilizing the COVID-19 death cases and consumer price index (CPI) for food at home as background information, we proposed an understanding of the pandemic’s impact on online grocery shopping by assembling, geovisualizing, and analyzing the evolution of online grocery behaviors and discussion on social media before and during the pandemic.

A global portrait of expressed mental health signals towards COVID-19 in social media space.

Globally, the COVID-19 pandemic has induced a mental health crisis. Social media data offer a unique opportunity to track the mental health signals of a given population and quantify their negativity towards COVID-19. To date, however, we know little about how negative sentiments differ across countries and how these relate to the shifting policy landscape experienced through the pandemic. Using 2.1 billion individual-level geotagged tweets posted between 1 February 2020 and 31 March 2021, we track, monitor and map the shifts in negativity across 217 countries and unpack its relationship with COVID-19 policies. Findings reveal that there are important geographic, demographic, and socioeconomic disparities of negativity across continents, different levels of a nation's income, population density, and the level of COVID-19 infection. Countries with more stringent policies were associated with lower levels of negativity, a relationship that weakened in later phases of the pandemic. This study provides the first global and multilingual evaluation of the public's real-time mental health signals to COVID-19 at a large spatial and temporal scale. We offer an empirical framework to monitor mental health signals globally, helping international authorizations, including the United Nations and World Health Organization, to design smart country-specific mental health initiatives in response to the ongoing pandemic and future public emergencies.