Geo-located Tweets Research Articles

Tweeting during the Pandemic in New York City: Unveiling the Evolving Sentiment Landscape of NYC through a Spatiotemporal Analysis of Geolocated Tweets

ABSTRACT This article explores the relationship between spatial factors, socioeconomic conditions, and Twitter (now called X) sentiment in New York City (NYC) during the COVID-19 pandemic. Using Twitter data, the study investigates how sentiment varied across different geographies. It examines whether sentiment scores, unemployment rates, and COVID-19 hospitalization rates in NYC zip codes revealed spatial associations. The research employs sentiment analysis, a natural language processing technique used to algorithmically determine the emotional tone of a text, on a database of geo-located tweets spanning January to December 2020. The findings reveal a shift towards more negative sentiment during the initial year of the pandemic. Moreover, the study uncovers variations in sentiment trends across boroughs and zip codes. Additionally, a zip code-level fixed-effects model demonstrates a statistically significant relationship between sentiment scores and unemployment rates. In summary, this article makes a two-fold contribution: firstly, it adds a spatial lens to the scholarly debate regarding the use of Twitter data as an indicator of publicly expressed sentiment; secondly, it provides empirical evidence on the spatial interconnectedness of sentiment, health (hospitalization), and socioeconomic factors (unemployment). Overall, this research sheds light on the nuanced relationship between sentiment and space during the COVID-19 pandemic in NYC.

Publicly expressed climate scepticism is greatest in regions with high CO2 emissions

We analysed a recently released corpus of climate-related tweets to examine the macro-level factors associated with public declarations of climate change scepticism. Analyses of over 2 million geo-located tweets in the U.S. showed that climate scepticism – and the aggressiveness of climate-related tweets – was greater in states with higher per capita carbon emissions. This pattern remained significant after controlling for political conservatism, GDP per capita, education, and gender, and was replicated across 126 nations from around the world. The findings are consistent with a vested interest hypothesis—misinformation around climate change is most likely to be distributed in regions where there is high fossil fuel reliance, and where the economic stakes of acknowledging climate change are high. Understanding the macro-level patterns that are implicated in climate scepticism can help inform structural interventions for those seeking a low-carbon future.

The migrant perspective: Measuring migrants' movements and interests using geolocated tweets

AbstractGeolocated social media data hold a hitherto untapped potential for exploring the relationship between user mobility and their interests at a large scale. Using geolocated Twitter data from Nigeria, we provide a feasibility study that demonstrates how the linkage of (1) a trajectory analysis of Twitter users' geolocation and (2) natural language processing of Twitter users' text content can reveal information about the interests of migrants. After identifying migrants via a trajectory analysis, we train a language model to automatically detect the topics of the migrants' tweets. Biases of manual labelling are circumvented by learning community‐defined topics from a Nigerian web forum. Results suggest that differences in users' mobility correlate with varying interests in several topics, most notably religion. We find that Twitter data can be a flexible source for exploring the link between users' mobility and interests in large‐scale analyses of urban populations. The joint use of spatial techniques and text analysis enables migration researchers to (a) study migrant perspectives in greater detail than is possible with census data and (b) at a larger scale than is feasible with interviews. Thereby, it provides a valuable complement to interviews, surveys and censuses, and holds a large potential for further research.

Visualizing 2018 lombok earthquake in Indonesia using crowdsouring data

Along with the development of science and technology, using big data, map makers can take advantage of crowdsourcing social media data on Twitter to obtain user location when uploading tweets, which can be called geolocated tweets. Earthquakes that occur very often in Indonesia often grab people's attention, especially netizens who use social media like Twitter. One of the major earthquakes that occurred in Indonesia in 2018 was the Lombok earthquake, which occurred twice in a row from July to August 2018. Using Twitter data, information and social responses related to the 2018 Lombok earthquake can be obtained, which can be used as evaluation material for public handling and responding. The information is then visualized in various forms, and one of the best visualization methods is selected.This study uses Twint package in Python as a way of obtaining location data from Twitter. The method used to collect Twitter data is a case study on the social impact of the Lombok earthquake in Indonesia in 2018. The data observation method used is a simulation of several types of map visualization and survey methods in selecting the best type of visualization. The method of analysis used is by mapping the data on the number of tweets as the main object using various types of maps, as well as calculating survey results by scoring each group of questions.The results of spatial data extraction from Twitter in this study obtained 2032 tweets that had been selected and cleaned from 11,584 tweets. Map visualization with the theme of the social impact of the Lombok earthquake in 2018 was compiled using five types of visualization, namely choropleth maps, proportional symbol maps, dot maps, hexagonal tessellation maps, and heat maps. Based on the results of the survey on selecting the best visualization, it was found that the choropleth map is the best visualization method according to respondents with a cartography background and respondents who are unfamiliar with cartography because the information displayed is easier to read and understand.

Leveraging Multiyear, Geospatial Social Media Data for Health Policy Evaluations: Lessons From the Philadelphia Beverage Tax.

Public reactions to health policies are vital to understand policy sustainability and impact but have been elusively difficult to dynamically measure. The 2021 launch of the Twitter Academic Application Programming Interface (API), allowing for historical tweet analyses, represents a potentially powerful tool for complex, comprehensive policy analyses. Using the Philadelphia Beverage Tax (implemented January 2017) as a case study, this research extracted longitudinal and geographic changes in sentiments, and key influencers in policy-related conversations. The Twitter API was used to retrieve all publicly available tweets related to the Tax between 2016 and 2019. Twitter. Users who posted publicly available tweets related to the Philadelphia Beverage Tax (PBT). Tweet content, frequency, sentiment, and user-related information. Tweet content, authors, engagement, and location were analyzed in parallel to key PBT events. Published emotional lexicons were used for sentiment analyses. A total of 45 891 tweets were retrieved (1311 with geolocation data). Changes in the tweet volume and sentiment were strongly driven by Tax-related litigation. While anger and fear increased in the months prior to the policy's implementation, they progressively decreased after its implementation; trust displayed an inverse trend. The 50 tweeters with the highest positive engagement included media outlets (n = 24), displaying particularly high tweet volume/engagement, and public personalities (n = 10), displaying the greatest polarization in tweet sentiment. Most geo-located tweets, reflecting 321 unique locations, were from the Philadelphia region (55.2%). Sentiment and positive engagement varied, although concentrations of negative sentiments were observed in some Philadelphia suburbs. Findings highlighted how longitudinal Twitter data can be leveraged to deconstruct specific, dynamic insights on public policy reactions and information dissemination to inform better policy implementation and evaluation (eg, anticipating catalysts for both heightened public interest and geographic, sentiment changes in policy conversations). This study provides policymakers a blueprint to conduct similar cost and time efficient yet dynamic and multifaceted health policy evaluations.

A tale of three cities: uncovering human-urban interactions with geographic-context aware social media data

Seeking spatiotemporal patterns about how citizens interact with the urban space is critical for understanding how cities function. Such interactions were studied in various forms focusing on patterns of people’s presence, action, and transition in the urban environment, which are defined as human-urban interactions in this paper. Using human activity datasets that utilize mobile positioning technology for tracking the locations and movements of individuals, researchers developed stochastic models to uncover preferential return behaviors and recurrent transitional activity structures in human-urban interactions. Ad-hoc heuristics and spatial clustering methods were applied to derive meaningful activity places in those studies. However, the lack of semantic meaning in the recorded locations makes it difficult to examine the details about how people interact with different activity places. In this study, we utilized geographic context-aware Twitter data to investigate the spatiotemporal patterns of people’s interactions with their activity places in different urban settings. To test consistency of our findings, we used geo-located tweets to derive the activity places in Twitter users’ location histories over three major U.S. metropolitan areas: Greater Boston Area, Chicago, and San Diego, where the geographic context of each location was inferred from its closest land use parcel. The results showed striking spatial and temporal similarities in Twitter users’ interactions with their activity places among the three cities. By using entropy-based predictability measures, this study not only confirmed the preferential return behaviors as people tend to revisit a few highly frequented places but also revealed detailed characteristics of those activity places.

OP74 Assessing Public Confidence Towards COVID-19 Vaccines Through Social Media Insights Leveraged Using Artificial Intelligence Techniques

IntroductionIn areas where public confidence is low and there is a lack of understanding around behaviors, such as COVID-19 vaccine hesitancy, there is a need to explore novel sources of evidence. When leveraged using artificial intelligence (AI) techniques, social media data may offer rich insights into public concerns around vaccination. Currently, sources of ‘soft-intelligence’ are underutilized by policy makers, health technology assessment (HTA) and other public health research agencies. In this work, we used an AI platform to rapidly detect and analyze key barriers to vaccine uptake from a sample of geo-located tweets.MethodsAn AI-based tool was deployed using a robust search strategy to capture tweets associated with COVID-19 vaccination, posted from users in London, United Kingdom. The tool’s algorithm automatically clustered tweets based on key topics of discussion and sentiment. Tweets contained within the 12 most populated topics with negative sentiment were extracted. The extracted tweets were mapped to one of six pre-determined themes (safety, mistrust, under-representation, complacency, ineffectiveness, and access) informed using the World Health Organization’s 3Cs vaccine hesitancy model. All collated tweets were anonymized.ResultsWe identified 91,473 tweets posted between 30 November 2020 and 15 August 2021. A sample of 913 tweets were extracted from the twelve negative topic clusters. Of these, 302 tweets were coded to a vaccine hesitancy theme. ‘Safety’ (29%) and ‘mistrust’ (23%) were the most commonly coded themes; the least commonly coded was ‘under-representation’ (3%). Within the main themes, adverse reactions, inadequate assessment, and rushed development of the vaccines as key findings. Our analysis also revealed widespread sharing of misinformation.ConclusionsUsing an AI-based text analytics tool, we were able to rapidly assess public confidence in COVID-19 vaccination and identify key barriers to uptake from a corpus of geo-located tweets. Our findings support a growing body of evidence and confidence surrounding the use of AI tools to efficiently analyze early sources of soft-intelligence evidence in public health research.

Using Social Media to Predict Food Deserts in the United States: Infodemiology Study of Tweets

BackgroundThe issue of food insecurity is becoming increasingly important to public health practitioners because of the adverse health outcomes and underlying racial disparities associated with insufficient access to healthy foods. Prior research has used data sources such as surveys, geographic information systems, and food store assessments to identify regions classified as food deserts but perhaps the individuals in these regions unknowingly provide their own accounts of food consumption and food insecurity through social media. Social media data have proved useful in answering questions related to public health; therefore, these data are a rich source for identifying food deserts in the United States.ObjectiveThe aim of this study was to develop, from geotagged Twitter data, a predictive model for the identification of food deserts in the United States using the linguistic constructs found in food-related tweets.MethodsTwitter’s streaming application programming interface was used to collect a random 1% sample of public geolocated tweets across 25 major cities from March 2020 to December 2020. A total of 60,174 geolocated food-related tweets were collected across the 25 cities. Each geolocated tweet was mapped to its respective census tract using point-to-polygon mapping, which allowed us to develop census tract–level features derived from the linguistic constructs found in food-related tweets, such as tweet sentiment and average nutritional value of foods mentioned in the tweets. These features were then used to examine the associations between food desert status and the food ingestion language and sentiment of tweets in a census tract and to determine whether food-related tweets can be used to infer census tract–level food desert status.ResultsWe found associations between a census tract being classified as a food desert and an increase in the number of tweets in a census tract that mentioned unhealthy foods (P=.03), including foods high in cholesterol (P=.02) or low in key nutrients such as potassium (P=.01). We also found an association between a census tract being classified as a food desert and an increase in the proportion of tweets that mentioned healthy foods (P=.03) and fast-food restaurants (P=.01) with positive sentiment. In addition, we found that including food ingestion language derived from tweets in classification models that predict food desert status improves model performance compared with baseline models that only include socioeconomic characteristics.ConclusionsSocial media data have been increasingly used to answer questions related to health and well-being. Using Twitter data, we found that food-related tweets can be used to develop models for predicting census tract food desert status with high accuracy and improve over baseline models. Food ingestion language found in tweets, such as census tract–level measures of food sentiment and healthiness, are associated with census tract–level food desert status.

An evaluation of geo-located Twitter data for measuring human migration

This study evaluates the spatial patterns of flows generated from geo-located Twitter data to measure human migration. Using geo-located tweets continuously collected in the U.S. from 2013 to 2015, we identified Twitter users who migrated per changes in county-of-residence every two years and compared the Twitter-estimated county-to-county migration flows with the ones from the U.S. Internal Revenue Service (IRS). To evaluate the spatial patterns of Twitter migration flows when representing the IRS counterparts, we developed a normalized difference representation index to visualize and identify those counties of over-/under-representations in the Twitter estimates. Further, we applied a multidimensional spatial scan statistic approach based on a Poisson process model to detect pairs of origin and destination regions where the over-/under-representativeness occurred. The results suggest that Twitter migration flows tend to under-represent the IRS estimates in regions with a large population and over-represent them in metropolitan regions adjacent to tourist attractions. This study demonstrated that geo-located Twitter data could be a sound statistical proxy for measuring human migration. Given that the spatial patterns of Twitter-estimated migration flows vary significantly across the geographic space, related studies will benefit from our approach by identifying those regions where data calibration is necessary.

Spatial Distribution of Hateful Tweets Against Asians and Asian Americans During the COVID-19 Pandemic, November 2019 to May 2020.

Objectives. To illustrate the spatiotemporal distribution of geolocated tweets that contain anti-Asian hate language in the contiguous United States during the early phase of the COVID-19 pandemic. Methods. We used a data set of geolocated tweets that match with keywords reflecting COVID-19 and anti-Asian hate and identified geographical clusters using the space-time scan statistic with Bernoulli model. Results. Anti-Asian hate language surged between January and March 2020. We found clusters of hate across the contiguous United States. The strongest cluster consisted of a single county (Ross County, Ohio), where the proportion of hateful tweets was 312.13 times higher than for the rest of the country. Conclusions. Anti-Asian hate on Twitter exhibits a significantly clustered spatiotemporal distribution. Clusters vary in size, duration, strength, and location and are scattered across the entire contiguous United States. Public Health Implications. Our results can inform decision-makers in public health and safety for allocating resources for place-based preparedness and response for pandemic-induced racism as a public health threat. (Am J Public Health. 2022;112(4):646-649. https://doi.org/10.2105/AJPH.2021.306653.

COVID-19 Vaccine Hesitancy: Analysing Twitter to Identify Barriers to Vaccination in a Low Uptake Region of the UK.

To facilitate effective targeted COVID-19 vaccination strategies, it is important to understand reasons for vaccine hesitancy where uptake is low. Artificial intelligence (AI) techniques offer an opportunity for real-time analysis of public attitudes, sentiments, and key discussion topics from sources of soft-intelligence, including social media data. In this work, we explore the value of soft-intelligence, leveraged using AI, as an evidence source to support public health research. As a case study, we deployed a natural language processing (NLP) platform to rapidly identify and analyse key barriers to vaccine uptake from a collection of geo-located tweets from London, UK. We developed a search strategy to capture COVID-19 vaccine related tweets, identifying 91,473 tweets between 30 November 2020 and 15 August 2021. The platform's algorithm clustered tweets according to their topic and sentiment, from which we extracted 913 tweets from the top 12 negative sentiment topic clusters. These tweets were extracted for further qualitative analysis. We identified safety concerns; mistrust of government and pharmaceutical companies; and accessibility issues as key barriers limiting vaccine uptake. Our analysis also revealed widespread sharing of vaccine misinformation amongst Twitter users. This study further demonstrates that there is promising utility for using off-the-shelf NLP tools to leverage insights from social media data to support public health research. Future work to examine where this type of work might be integrated as part of a mixed-methods research approach to support local and national decision making is suggested.

Real-world trajectory sharing with local differential privacy

Sharing trajectories is beneficial for many real-world applications, such as managing disease spread through contact tracing and tailoring public services to a population's travel patterns. However, public concern over privacy and data protection has limited the extent to which this data is shared. Local differential privacy enables data sharing in which users share a perturbed version of their data, but existing mechanisms fail to incorporate user-independent public knowledge (e.g., business locations and opening times, public transport schedules, geo-located tweets). This limitation makes mechanisms too restrictive, gives unrealistic outputs, and ultimately leads to low practical utility. To address these concerns, we propose a local differentially private mechanism that is based on perturbing hierarchically-structured, overlapping n -grams (i.e., contiguous subsequences of length n ) of trajectory data. Our mechanism uses a multi-dimensional hierarchy over publicly available external knowledge of real-world places of interest to improve the realism and utility of the perturbed, shared trajectories. Importantly, including real-world public data does not negatively affect privacy or efficiency. Our experiments, using real-world data and a range of queries, each with real-world application analogues, demonstrate the superiority of our approach over a range of alternative methods.

Achieving situational awareness of drug cartels with geolocated social media

Using geolocated tweets to achieve situational awareness is an often researched topic in disaster and emergency management. However, little has been done in the area of drug cartels, which, as transnational crime organizations, continue to pose great risk to the stability and safety of our communities. This paper made an initial effort in using geolocated social media (specifically Twitter) to achieve situational awareness of drug cartels through temporal and spatial analysis of derived named entity clusters. The results show that detecting peaks in the time series of frequently occurring entity clusters enabled the tracking of important events in public discourse surrounding drug cartels. Correlations between time series also provided valuable insights into the synchronicity between different events. Further examining the spatial distribution of key events for different countries, we identify thematic hotpots of public discourse on cartel activity. Our methodology also addresses issues of language ambiguity when working with noisy social media data in order to achieve situational awareness on drug cartels.Please check and confirm the city name is correctly identified in affiliations 1 and 2.confirmedPlease check and confirm the corresponding author and mail ID is correctly identified.confirmed

From Data to Rhizomes: Applying a Geographical Concept to Understand the Mobility of Tourists from Geo-Located Tweets

In geography, the concept of “rhizome” provides a theoretical tool to conceive the way people move in space in terms of “mobility networks”: the space lived by people is delimited and characterized on the basis of both the places they visited and the sequences of their transfers from place to place. Researchers are now wondering whether in the new era of data-driven geography it is possible to give a concrete shape to the concept of rhizome, by analyzing big data describing movement of people traced through social media. This paper is a first attempt to give a concrete shape to the concept of rhizome, by interpreting it as a problem of “itemset mining”, which is a well-known data mining technique. This technique was originally developed for market-basket analysis. We studied how the application of this technique, if supported by adequate visualization strategies, can provide geographers with a concrete shape for rhizomes, suitable for further studies. To validate the ideas, we chose the case study of tourists visiting a city: the rhizome can be conceived as the set of places visited by many tourists, and the common transfers made by tourists in the area of the city. Itemsets extracted from a real-life data set were used to study the effectiveness of both a topographic representation and a topological representation to visualize rhizomes. In this paper, we study how three different interpretations are actually able to give a concrete and visual shape to the concept of rhizome. The results that we present and discuss in this paper open further investigations on the problem.

From twitter to GDP: Estimating economic activity from social media

Using all geo-located image tweets shared on Twitter in 2012–2013, I find that the volume of tweets is a valid proxy for estimating GDP at the country level, explaining 78 percent of cross-country variations. I also exploit the geographic granularity of social media posts to estimate and predict GDP at the sub-national level. I find that tweets alone can explain 52 percent of the variation in GDP across cities in the US. Estimates using Twitter data perform on par with the more common night-lights proxy. Furthermore, both indicators seem to capture different aspects of economic activity and thus complement each other.

News and Geolocated Social Media Accurately Measure Protest Size Variation

Larger protests are more likely to lead to policy changes than small ones are, but whether or not attendance estimates provided in news or generated from social media are biased is an open question. This letter closes the question: news and geolocated social media data generate accurate estimates of protest size variation. This claim is substantiated using cellphone location data from more than 10 million individuals during the 2017 United States Women’s March protests. These cellphone estimates correlate strongly with those provided in news media as well as three size estimates generated using geolocated tweets, one text-based and two based on images. Inferences about protest attendance from these estimates match others’ findings about the Women’s March.

The geography of sentiment towards the Women's March of 2017.

The Women’s March of 2017 generated unprecedented levels of participation in the largest, single day, protest in history to date. The marchers protested the election of President Donald Trump and rallied in support of several civil issues such as women’s rights. “Sister marches” evolved in at least 680 locations across the United States. Both positive and negative reactions to the March found their way into social media, with criticism stemming from certain, conservative, political sources and other groups. In this study, we investigate the extent to which this notable, historic event influenced sentiment on Twitter, and the degree to which responses differed by geographic area within the continental U.S. Tweets about the event rose to an impressive peak of over 12% of all geo-located tweets by mid-day of the March, Jan. 21. Messages included in tweets associated with the March tended to be positive in sentiment, on average, with a mean of 0.34 and a median of 0.07 on a scale of -4 to +4. In fact, tweets associated with the March were more positive than all other geo-located tweets during the day of the March. Exceptions to this pattern of positive sentiment occurred only in seven metropolitan areas, most of which involved very small numbers of tweets. Little evidence surfaced of extensive patterns of negative, aggressive messages towards the event in this set of tweets. Given the widespread nature of online harassment and sexist tweets, more generally, the results are notable. In sum, online reactions to the March on this social media platform suggest that this modern arm of the Women’s Movement received considerable, virtual support across the country.

A Statistical Approach for Studying the Spatio-Temporal Distribution of Geolocated Tweets in Urban Environments

An in-depth descriptive approach to the dynamics of the urban population is fundamental as a first step towards promoting effective planning and designing processes in cities. Understanding the behavioral aspects of human activities can contribute to their effective management and control. We present a framework, based on statistical methods, for studying the spatio-temporal distribution of geolocated tweets as a proxy for where and when people carry out their activities. We have evaluated our proposal by analyzing the distribution of collected geolocated tweets over a two-week period in the summer of 2017 in Lisbon, London, and Manhattan. Our proposal considers a negative binomial regression analysis for the time series of counts of tweets as a first step. We further estimate a functional principal component analysis of second-order summary statistics of the hourly spatial point patterns formed by the locations of the tweets. Finally, we find groups of hours with a similar spatial arrangement of places where humans develop their activities through hierarchical clustering over the principal scores. Social media events are found to show strong temporal trends such as seasonal variation due to the hour of the day and the day of the week in addition to autoregressive schemas. We have also identified spatio-temporal patterns of clustering, i.e., groups of hours of the day that present a similar spatial distribution of human activities.

Spatio-Temporal Distribution of Negative Emotions in New York City After a Natural Disaster as Seen in Social Media.

Disasters have substantial consequences for population mental health. We used Twitter to (1) extract negative emotions indicating discomfort in New York City (NYC) before, during, and after Superstorm Sandy in 2012. We further aimed to (2) identify whether pre- or peri-disaster discomfort were associated with peri- or post-disaster discomfort, respectively, and to (3) assess geographic variation in discomfort across NYC census tracts over time. Our sample consisted of 1,018,140 geo-located tweets that were analyzed with an advanced sentiment analysis called ”Extracting the Meaning Of Terse Information in a Visualization of Emotion” (EMOTIVE). We calculated discomfort rates for 2137 NYC census tracts, applied spatial regimes regression to find associations of discomfort, and used Moran’s I for spatial cluster detection across NYC boroughs over time. We found increased discomfort, that is, bundled negative emotions after the storm as compared to during the storm. Furthermore, pre- and peri-disaster discomfort was positively associated with post-disaster discomfort; however, this association was different across boroughs, with significant associations only in Manhattan, the Bronx, and Queens. In addition, rates were most prominently spatially clustered in Staten Island lasting pre- to post-disaster. This is the first study that determined significant associations of negative emotional responses found in social media posts over space and time in the context of a natural disaster, which may guide us in identifying those areas and populations mostly in need for care.

Opportunity makes opinion leaders: analyzing the role of first-hand information in opinion leadership in social media networks

ABSTRACTTheorizing information flows is at the heart of traditional communication theories such as the two-step flow of communication and the concept of opinion leadership. Social media have fundamentally altered how information reaches people. This study examines opinion leadership in social media networks and argues that opinion leaders may no longer need to rely on information provided by the media if they have access to first-hand information. To test this assumption empirically, we used data from the 2015 United Nations Climate Change Conference (COP21). Attendees of the conference had direct information about what was happening, which they were able to share live with their followers via social media. We used geo-located tweets to identify Twitter users who attended the COP21 summit. We then located these users in a data set of tweets that were collected based on the main conference hashtag (#COP21) and represent the wider social media debate on the conference. Our results, which are based on network analysis measures and Twitter user data, show that COP21 participants were more central actors compared to the average user in the network, and that they were more likely to have brokering positions. They were also more involved in the debate and received more attention from other users. We used automated content analysis to divide COP21 participants into different actor types and performed the analysis by actor group. The results show only minor differences across the actors and support the robustness of our analysis.