The New York Times is releasing a series of data files with cumulative counts of coronavirus cases in the United States, at the state and county level, over time. We are compiling this time series data from state and local governments and health departments in an attempt to provide a complete record of the ongoing outbreak.

Since late January, The Times has tracked cases of coronavirus in real time as they were identified after testing. Because of the widespread shortage of testing, however, the data is necessarily limited in the picture it presents of the outbreak.

We have used this data to power our maps and reporting tracking the outbreak, and it is now being made available to the public in response to requests from researchers, scientists and government officials who would like access to the data to better understand the outbreak.

The data begins with the first reported coronavirus case in Washington State on Jan. 21, 2020. We will publish regular updates to the data in this repository.

The COVID Tracking Project collects information from 50 US states, the District of Columbia, and 5 other US territories to provide the most comprehensive testing data we can collect for the novel coronavirus, SARS-CoV-2. We attempt to include positive and negative results, pending tests, and total people tested for each state or district currently reporting that data.

Testing is a crucial part of any public health response, and sharing test data is essential to understanding this outbreak. The CDC is currently not publishing complete testing data, so we’re doing our best to collect it from each state and provide it to the public. The information is patchy and inconsistent, so we’re being transparent about what we find and how we handle it—the spreadsheet includes our live comments about changing data and how we’re working with incomplete information.

From here, you can also learn about our methodology, see who makes this, and find out what information states provide and how we handle it.

Introduction

There are several works based on Natural Language Processing on newspaper reports. Mining opinions from headlines [ 1 ] using Standford NLP and SVM by Rameshbhaiet. Al.compared several algorithms on a small and large dataset. Rubinet. al., in their paper [ 2 ], created a mechanism to differentiate fake news from real ones by building a set of characteristics of news according to their types. The purpose was to contribute to the low resource data available for training machine learning algorithms. Doumitet. al.in [ 3 ] have implemented LDA, a topic modeling approach to study bias present in online news media.

However, there are not many NLP research invested in studying COVID-19. Most applications include classification of chest X-rays and CT-scans to detect presence of pneumonia in lungs [ 4 ], a consequence of the virus. Other research areas include studying the genome sequence of the virus[ 5 ][ 6 ][ 7 ] and replicating its structure to fight and find a vaccine. This research is crucial in battling the pandemic. The few NLP based research publications are sentiment classification of online tweets by Samuel et el [ 8 ] to understand fear persisting in people due to the virus. Similar work has been done using the LSTM network to classify sentiments from online discussion forums by Jelodaret. al.[ 9 ]. NKK dataset is the first study on a comparatively larger dataset of a newspaper report on COVID-19, which contributed to the virus’s awareness to the best of our knowledge.

2 Data-set Introduction

2.1 Data Collection

We accumulated 1000 online newspaper report from United States of America (USA) on COVID-19. The newspaper includes The Washington Post (USA) and StarTribune (USA). We have named it as “Covid-News-USA-NNK”. We also accumulated 50 online newspaper report from Bangladesh on the issue and named it “Covid-News-BD-NNK”. The newspaper includes The Daily Star (BD) and Prothom Alo (BD). All these newspapers are from the top provider and top read in the respective countries. The collection was done manually by 10 human data-collectors of age group 23- with university degrees. This approach was suitable compared to automation to ensure the news were highly relevant to the subject. The newspaper online sites had dynamic content with advertisements in no particular order. Therefore there were high chances of online scrappers to collect inaccurate news reports. One of the challenges while collecting the data is the requirement of subscription. Each newspaper required $1 per subscriptions. Some criteria in collecting the news reports provided as guideline to the human data-collectors were as follows:

The headline must have one or more words directly or indirectly related to COVID-19.

The content of each news must have 5 or more keywords directly or indirectly related to COVID-19.

The genre of the news can be anything as long as it is relevant to the topic. Political, social, economical genres are to be more prioritized.

Avoid taking duplicate reports.

Maintain a time frame for the above mentioned newspapers.

To collect these data we used a google form for USA and BD. We have two human editor to go through each entry to check any spam or troll entry.

2.2 Data Pre-processing and Statistics

Some pre-processing steps performed on the newspaper report dataset are as follows:

Remove hyperlinks.

Remove non-English alphanumeric characters.

Remove stop words.

Lemmatize text.

While more pre-processing could have been applied, we tried to keep the data as much unchanged as possible since changing sentence structures could result us in valuable information loss. While this was done with help of a script, we also assigned same human collectors to cross check for any presence of the above mentioned criteria.

The primary data statistics of the two dataset are shown in Table 1 and 2.

Table 1: Covid-News-USA-NNK data statistics

No of words per headline

7 to 20

No of words per body content

150 to 2100

Table 2: Covid-News-BD-NNK data statistics No of words per headline

10 to 20

No of words per body content

100 to 1500

2.3 Dataset Repository

We used GitHub as our primary data repository in account name NKK^1. Here, we created two repositories USA-NKK^2 and BD-NNK^3. The dataset is available in both CSV and JSON format. We are regularly updating the CSV files and regenerating JSON using a py script. We provided a python script file for essential operation. We welcome all outside collaboration to enrich the dataset.

3 Literature Review

Natural Language Processing (NLP) deals with text (also known as categorical) data in computer science, utilizing numerous diverse methods like one-hot encoding, word embedding, etc., that transform text to machine language, which can be fed to multiple machine learning and deep learning algorithms.

Some well-known applications of NLP includes fraud detection on online media sites[ 10 ], using authorship attribution in fallback authentication systems[ 11 ], intelligent conversational agents or chatbots[ 12 ] and machine translations used by Google Translate[ 13 ]. While these are all downstream tasks, several exciting developments have been made in the algorithm solely for Natural Language Processing tasks. The two most trending ones are BERT[ 14 ], which uses bidirectional encoder-decoder architecture to create the transformer model, that can do near-perfect classification tasks and next-word predictions for next generations, and GPT-3 models released by OpenAI[ 15 ] that can generate texts almost human-like. However, these are all pre-trained models since they carry huge computation cost. Information Extraction is a generalized concept of retrieving information from a dataset. Information extraction from an image could be retrieving vital feature spaces or targeted portions of an image; information extraction from speech could be retrieving information about names, places, etc[ 16 ]. Information extraction in texts could be identifying named entities and locations or essential data. Topic modeling is a sub-task of NLP and also a process of information extraction. It clusters words and phrases of the same context together into groups. Topic modeling is an unsupervised learning method that gives us a brief idea about a set of text. One commonly used topic modeling is Latent Dirichlet Allocation or LDA[17].

Keyword extraction is a process of information extraction and sub-task of NLP to extract essential words and phrases from a text. TextRank [ 18 ] is an efficient keyword extraction technique that uses graphs to calculate the weight of each word and pick the words with more weight to it.

Word clouds are a great visualization technique to understand the overall ’talk of the topic’. The clustered words give us a quick understanding of the content.

4 Our experiments and Result analysis

We used the wordcloud library^4 to create the word clouds. Figure 1 and 3 presents the word cloud of Covid-News-USA- NNK dataset by month from February to May. From the figures 1,2,3, we can point few information:

In February, both the news paper have talked about China and source of the outbreak.

StarTribune emphasized on Minnesota as the most concerned state. In April, it seemed to have been concerned more.

Both the newspaper talked about the virus impacting the economy, i.e, bank, elections, administrations, markets.

Washington Post discussed global issues more than StarTribune.

StarTribune in February mentioned the first precautionary measurement: wearing masks, and the uncontrollable spread of the virus throughout the nation.

While both the newspaper mentioned the outbreak in China in February, the weight of the spread in the United States are more highlighted through out March till May, displaying the critical impact caused by the virus.

We used a script to extract all numbers related to certain keywords like ’Deaths’, ’Infected’, ’Died’ , ’Infections’, ’Quarantined’, Lock-down’, ’Diagnosed’ etc from the news reports and created a number of cases for both the newspaper. Figure 4 shows the statistics of this series. From this extraction technique, we can observe that April was the peak month for the covid cases as it gradually rose from February. Both the newspaper clearly shows us that the rise in covid cases from February to March was slower than the rise from March to April. This is an important indicator of possible recklessness in preparations to battle the virus. However, the steep fall from April to May also shows the positive response against the attack. We used Vader Sentiment Analysis to extract sentiment of the headlines and the body. On average, the sentiments were from -0.5 to -0.9. Vader Sentiment scale ranges from -1(highly negative to 1(highly positive). There were some cases

where the sentiment scores of the headline and body contradicted each other,i.e., the sentiment of the headline was negative but the sentiment of the body was slightly positive. Overall, sentiment analysis can assist us sort the most concerning (most negative) news from the positive ones, from which we can learn more about the indicators related to COVID-19 and the serious impact caused by it. Moreover, sentiment analysis can also provide us information about how a state or country is reacting to the pandemic. We used PageRank algorithm to extract keywords from headlines as well as the body content. PageRank efficiently highlights important relevant keywords in the text. Some frequently occurring important keywords extracted from both the datasets are: ’China’, Government’, ’Masks’, ’Economy’, ’Crisis’, ’Theft’ , ’Stock market’ , ’Jobs’ , ’Election’, ’Missteps’, ’Health’, ’Response’. Keywords extraction acts as a filter allowing quick searches for indicators in case of locating situations of the economy,

Updates

• Notice of data discontinuation: Since the start of the pandemic, AP has reported case and death counts from data provided by Johns Hopkins University. Johns Hopkins University has announced that they will stop their daily data collection efforts after March 10. As Johns Hopkins stops providing data, the AP will also stop collecting daily numbers for COVID cases and deaths. The HHS and CDC now collect and visualize key metrics for the pandemic. AP advises using those resources when reporting on the pandemic going forward.

  • CDC Weekly case and death counts (national and state level)
  • CDC County level cases and deaths
  • HHS New hospital admissions
  • CDC NowCast COVID variant proportions (national and regional level)

• April 9, 2020

  • The population estimate data for New York County, NY has been updated to include all five New York City counties (Kings County, Queens County, Bronx County, Richmond County and New York County). This has been done to match the Johns Hopkins COVID-19 data, which aggregates counts for the five New York City counties to New York County.

• April 20, 2020

  • Johns Hopkins death totals in the US now include confirmed and probable deaths in accordance with CDC guidelines as of April 14. One significant result of this change was an increase of more than 3,700 deaths in the New York City count. This change will likely result in increases for death counts elsewhere as well. The AP does not alter the Johns Hopkins source data, so probable deaths are included in this dataset as well.

• April 29, 2020

  • The AP is now providing timeseries data for counts of COVID-19 cases and deaths. The raw counts are provided here unaltered, along with a population column with Census ACS-5 estimates and calculated daily case and death rates per 100,000 people. Please read the updated caveats section for more information.

• September 1st, 2020

  • Johns Hopkins is now providing counts for the five New York City counties individually.

• February 12, 2021

  • The Ohio Department of Health recently announced that as many as 4,000 COVID-19 deaths may have been underreported through the state’s reporting system, and that the "daily reported death counts will be high for a two to three-day period."
  • Because deaths data will be anomalous for consecutive days, we have chosen to freeze Ohio's rolling average for daily deaths at the last valid measure until Johns Hopkins is able to back-distribute the data. The raw daily death counts, as reported by Johns Hopkins and including the backlogged death data, will still be present in the new_deaths column.

• February 16, 2021

  - Johns Hopkins has reconciled Ohio's historical deaths data with the state.

  Overview

The AP is using data collected by the Johns Hopkins University Center for Systems Science and Engineering as our source for outbreak caseloads and death counts for the United States and globally.

The Hopkins data is available at the county level in the United States. The AP has paired this data with population figures and county rural/urban designations, and has calculated caseload and death rates per 100,000 people. Be aware that caseloads may reflect the availability of tests -- and the ability to turn around test results quickly -- rather than actual disease spread or true infection rates.

This data is from the Hopkins dashboard that is updated regularly throughout the day. Like all organizations dealing with data, Hopkins is constantly refining and cleaning up their feed, so there may be brief moments where data does not appear correctly. At this link, you’ll find the Hopkins daily data reports, and a clean version of their feed.

The AP is updating this dataset hourly at 45 minutes past the hour.

To learn more about AP's data journalism capabilities for publishers, corporations and financial institutions, go here or email kromano@ap.org.

Queries

Use AP's queries to filter the data or to join to other datasets we've made available to help cover the coronavirus pandemic

• Filter cases by state here

• Rank states by their status as current hotspots. Calculates the 7-day rolling average of new cases per capita in each state: https://data.world/associatedpress/johns-hopkins-coronavirus-case-tracker/workspace/query?queryid=481e82a4-1b2f-41c2-9ea1-d91aa4b3b1ac

• Find recent hotspots within your state by running a query to calculate the 7-day rolling average of new cases by capita in each county: https://data.world/associatedpress/johns-hopkins-coronavirus-case-tracker/workspace/query?queryid=b566f1db-3231-40fe-8099-311909b7b687&showTemplatePreview=true

• Join county-level case data to an earlier dataset released by AP on local hospital capacity here. To find out more about the hospital capacity dataset, see the full details.

• Pull the 100 counties with the highest per-capita confirmed cases here

• Rank all the counties by the highest per-capita rate of new cases in the past 7 days here. Be aware that because this ranks per-capita caseloads, very small counties may rise to the very top, so take into account raw caseload figures as well.

Interactive

The AP has designed an interactive map to track COVID-19 cases reported by Johns Hopkins.

@(https://datawrapper.dwcdn.net/nRyaf/15/)

Interactive Embed Code

<iframe title="USA counties (2018) choropleth map Mapping COVID-19 cases by county" aria-describedby="" id="datawrapper-chart-nRyaf" src="https://datawrapper.dwcdn.net/nRyaf/10/" scrolling="no" frameborder="0" style="width: 0; min-width: 100% !important;" height="400"></iframe><script type="text/javascript">(function() {'use strict';window.addEventListener('message', function(event) {if (typeof event.data['datawrapper-height'] !== 'undefined') {for (var chartId in event.data['datawrapper-height']) {var iframe = document.getElementById('datawrapper-chart-' + chartId) || document.querySelector("iframe[src*='" + chartId + "']");if (!iframe) {continue;}iframe.style.height = event.data['datawrapper-height'][chartId] + 'px';}}});})();</script>

Caveats

• This data represents the number of cases and deaths reported by each state and has been collected by Johns Hopkins from a number of sources cited on their website.
• In some cases, deaths or cases of people who've crossed state lines -- either to receive treatment or because they became sick and couldn't return home while traveling -- are reported in a state they aren't currently in, because of state reporting rules.
• In some states, there are a number of cases not assigned to a specific county -- for those cases, the county name is "unassigned to a single county"
• This data should be credited to Johns Hopkins University's COVID-19 tracking project. The AP is simply making it available here for ease of use for reporters and members.
• Caseloads may reflect the availability of tests -- and the ability to turn around test results quickly -- rather than actual disease spread or true infection rates.
• Population estimates at the county level are drawn from 2014-18 5-year estimates from the American Community Survey.
• The Urban/Rural classification scheme is from the Center for Disease Control and Preventions's National Center for Health Statistics. It puts each county into one of six categories -- from Large Central Metro to Non-Core -- according to population and other characteristics. More details about the classifications can be found here.

Johns Hopkins timeseries data - Johns Hopkins pulls data regularly to update their dashboard. Once a day, around 8pm EDT, Johns Hopkins adds the counts for all areas they cover to the timeseries file. These counts are snapshots of the latest cumulative counts provided by the source on that day. This can lead to inconsistencies if a source updates their historical data for accuracy, either increasing or decreasing the latest cumulative count. - Johns Hopkins periodically edits their historical timeseries data for accuracy. They provide a file documenting all errors in their timeseries files that they have identified and fixed here

Attribution

This data should be credited to Johns Hopkins University COVID-19 tracking project

As the United States battles the coronavirus, news consumers across the country have been attempting to keep themselves updated with how the pandemic is progressing, and a survey held in March 2020 revealed that the most trusted news source for details on COVID-19 was the CDC, with ** percent of respondents saying that they trusted the centers to provide accurate information on the topic. Following closely behind was the World Health Organization and then the state government, but just ** percent of consumers said that they trusted social media sites to publish reliable and accurate news about the coronavirus outbreak.

Source

This data comes from the New York Times Coronavirus (Covid-19) Data in the United States GitHub repository. They use it to power their interactive page(s) on Covid-19, such as Coronavirus in the U.S.: Latest Map and Case Count.

What's Included?

The primary data published here are the daily cumulative number of cases and deaths reported in each county and state across the U.S. since the beginning of the pandemic. We have also published these additional data sets:

• Prisons: Cases in prisons
• Colleges: Cases on college and university campuses.
• Excess deaths: The elevated overall number of deaths during the pandemic.
• Mask use: A July 2020 survey of how regularly people in each county wore masks.
• Averages and anomalies: A set of pre-computed rolling averages of cases and deaths for ease of analysis or use in making graphics, along with a set of days with anomalous data that have been excluded from the averages.

The cumulative & rolling averages for cases and deaths are continually updated, but the more specific data mentioned above for prisons, etc. is no longer being updated.

This includes data at the national, state, and county levels.

License and Attribution

If you use this data, you must attribute it to “The New York Times” in any publication. If you would like a more expanded description of the data, you could say “Data from The New York Times, based on reports from state and local health agencies.”

Acknowledgements

Header Image: https://www.pexels.com/photo/n95-face-mask-3993241/

More Information

See the original New York Times source README which is also included in this dataset.

Context

This is the US Coronavirus data repository from The New York Times . This data includes COVID-19 cases and deaths reported by state and county. The New York Times compiled this data based on reports from state and local health agencies. More information on the data repository is available here . For additional reporting and data visualizations, see The New York Times’ U.S. coronavirus interactive site

Sample Queries

Query 1

Which US counties have the most confirmed cases per capita? This query determines which counties have the most cases per 100,000 residents. Note that this may differ from similar queries of other datasets because of differences in reporting lag, methodologies, or other dataset differences.

SELECT covid19.county, covid19.state_name, total_pop AS county_population, confirmed_cases, ROUND(confirmed_cases/total_pop *100000,2) AS confirmed_cases_per_100000, deaths, ROUND(deaths/total_pop *100000,2) AS deaths_per_100000 FROM bigquery-public-data.covid19_nyt.us_counties covid19 JOIN bigquery-public-data.census_bureau_acs.county_2017_5yr acs ON covid19.county_fips_code = acs.geo_id WHERE date = DATE_SUB(CURRENT_DATE(),INTERVAL 1 day) AND covid19.county_fips_code != "00000" ORDER BY confirmed_cases_per_100000 desc

Query 2

How do I calculate the number of new COVID-19 cases per day? This query determines the total number of new cases in each state for each day available in the dataset SELECT b.state_name, b.date, MAX(b.confirmed_cases - a.confirmed_cases) AS daily_confirmed_cases FROM (SELECT state_name AS state, state_fips_code , confirmed_cases, DATE_ADD(date, INTERVAL 1 day) AS date_shift FROM bigquery-public-data.covid19_nyt.us_states WHERE confirmed_cases + deaths > 0) a JOIN bigquery-public-data.covid19_nyt.us_states b ON a.state_fips_code = b.state_fips_code AND a.date_shift = b.date GROUP BY b.state_name, date ORDER BY date desc

Dataset Card for BEIR Benchmark

  Dataset Summary

BEIR is a heterogeneous benchmark that has been built from 18 diverse datasets representing 9 information retrieval tasks:

Fact-checking: FEVER, Climate-FEVER, SciFact Question-Answering: NQ, HotpotQA, FiQA-2018 Bio-Medical IR: TREC-COVID, BioASQ, NFCorpus News Retrieval: TREC-NEWS, Robust04 Argument Retrieval: Touche-2020, ArguAna Duplicate Question Retrieval: Quora, CqaDupstack Citation-Prediction: SCIDOCS Tweet… See the full description on the dataset page: https://huggingface.co/datasets/BeIR/trec-covid.

According to the most recently available data, around ********* of Americans feel very confident in their ability to check the accuracy of news stories regarding coronavirus. In an online survey conducted in **********, ** percent of respondents stated they would know how to confirm the accuracy of news and information regarding the COVID-19 pandemic. The majority of participants expressed a moderate level of self confidence in their capacity to fact check, with ** percent somewhat confident.

Constraint@AAAI2021 - COVID19 Fake News Detection in English @misc{patwa2020fighting, title={Fighting an Infodemic: COVID-19 Fake News Dataset}, author={Parth Patwa and Shivam Sharma and Srinivas PYKL and Vineeth Guptha and Gitanjali Kumari and Md Shad Akhtar and Asif Ekbal and Amitava Das and Tanmoy Chakraborty}, year={2020}, eprint={2011.03327}, archivePrefix={arXiv}, primaryClass={cs.CL} }

https://www.googleapis.com/download/storage/v1/b/kaggle-user-content/o/inbox%2F5505749%2F2b83271d61e47e2523e10dc9c28e545c%2F600x200.jpg?generation=1599042483103679&alt=media" alt="">

ALTADATA is a curated data marketplace where our subscribers and our data partners can easily exchange ready-to-analyze datasets and create insights with EPO, our visual data analytics platform.

COVID-19 US Daily Data

State level daily COVID-19 data for United States, provided by Johns Hopkins University (JHU) Center for Systems Science and Engineering (CSSE). If you want to use the updated version of the data, you can use our daily updated data with the help of api key by entering it via Altadata.

Overview

In this data product, you may find the latest and historical daily data on the COVID-19 pandemic for United States with the states level breakdown.

The COVID‑19 pandemic, also known as the coronavirus pandemic, is an ongoing global pandemic of coronavirus disease 2019 (COVID‑19), caused by severe acute respiratory syndrome coronavirus 2 (SARS‑CoV‑2). The outbreak was first identified in December 2019 in Wuhan, China. The World Health Organization declared the outbreak a Public Health Emergency of International Concern on 30 January 2020 and a pandemic on 11 March. As of 12 August 2020, more than 20.2 million cases of COVID‑19 have been reported in more than 188 countries and territories, resulting in more than 741,000 deaths; more than 12.5 million people have recovered.

The Johns Hopkins Coronavirus Resource Center is a continuously updated source of COVID-19 data and expert guidance. They aggregate and analyze the best data available on COVID-19 - including cases, as well as testing, contact tracing and vaccine efforts - to help the public, policymakers and healthcare professionals worldwide respond to the pandemic.

Methodology

• Cases and Death counts include confirmed and probable (where reported)
• Recovered cases are estimates based on local media reports, and state and local reporting when available, and therefore may be substantially lower than the true number. US state-level recovered cases are from COVID Tracking Project.
• Active cases = total cases - total recovered - total deaths
• Incidence Rate = cases per 100,000 persons
• Case-Fatality Ratio (%) = Number recorded deaths / Number cases
• US Testing Rate = total test results per 100,000 persons. The "total test results" are equal to "Total test results (Positive + Negative)" from COVID Tracking Project.
• US Hospitalization Rate (%) = Total number hospitalized / Number cases. The "Total number hospitalized" is the "Hospitalized – Cumulative" count from COVID Tracking Project. The "hospitalization rate" and "Total number hospitalized" are only presented for those states which provide cumulative hospital data.
• States Population data is retrieved from U.S. Census Bureau on top of the JHU CSSE's COVID-19 data

Data Source

• Johns Hopkins University Center for Systems Science and Engineering (JHU CSSE)
• U.S. Census Bureau, Population Division

Related Data Products

• COVID-19 Worldwide Daily Data

Suggested Blog Posts

• Bayesian Thinking During the Pandemic
• Impact of COVID-19 on California Electricity Demand
• Keep Calm and Look At The Fundamentals
• Markets In The Corona Virus Crisis

Data Dictionary

• Reported Date (reported_date): Covid-19 Report Date
• Province State (province_state): State name
• Population (population): Estimated state populations as of July 2019, as per U.S. Census Bureau Population Division
• Latitude (lat): Dot locations, not representative of a specific address
• Longitude (lng): Dot locations longitude, not representative of a specific address
• Confirmed Case (confirmed): Confirmed cases include presumptive positive cases and probable cases
• Active cases (active): Active cases = total confirmed - total recovered - total deaths
• Deaths (deaths): Death cases counts
• Recovered (recovered): Recovered cases counts
• Hospitalization Rate (hospitalization_rate): Total number of people hospitalized * 100...

A COVID-19 misinformation / fake news / rumor / disinformation dataset collected from online social media and news websites. Usage note:Misinformation detection, classification, tracking, prediction.Misinformation sentiment analysis.Rumor veracity classification, comment stance classification.Rumor tracking, social network analysis.Data pre-processing and data analysis codes available at https://github.com/MickeysClubhouse/COVID-19-rumor-datasetPlease see full info in our GitHub link.Cite us:Cheng, Mingxi, et al. "A COVID-19 Rumor Dataset." Frontiers in Psychology 12 (2021): 1566.@article{cheng2021covid, title={A COVID-19 Rumor Dataset}, author={Cheng, Mingxi and Wang, Songli and Yan, Xiaofeng and Yang, Tianqi and Wang, Wenshuo and Huang, Zehao and Xiao, Xiongye and Nazarian, Shahin and Bogdan, Paul}, journal={Frontiers in Psychology}, volume={12}, pages={1566}, year={2021}, publisher={Frontiers} }

According to a survey conducted in March 2020, ** percent of U.S. news consumers said that they were seeking out the latest information about the coronavirus via news media in general, including TV news, radio news, online news, and newspapers. In fact, ** percent of adults aged 55 or above were getting most of their news about the virus this way, compared to just ** percent of ** to 24-year-olds who were more likely than their older peers to turn to websites or social media posts from government or health agencies.

Dataset Card for COVID News Articles (2020 - 2022)

  Dataset Summary

The dataset encapsulates approximately half a million news articles collected over a period of 2 years during the Coronavirus pandemic onset and surge. It consists of 3 columns - title, content and category. title refers to the headline of the news article. content refers to the article in itself and category denotes the overall context of the news article at a high level. The dataset encapsulates… See the full description on the dataset page: https://huggingface.co/datasets/osanseviero/covid_news.

Project Overview This portion of the COVID DIARIES project provides full bibliographic information (including original and permanent links) to media items related to the COVID-19 vaccination program, published on the official websites of 20 major U.S. news outlets, including television networks, magazines, and newspapers. It spans the period from December 2020, when states began implementing Phase 1a of the vaccine allocation plan, through September 2021, when vaccines became widely available to all adults and were frequently mandated. News items were collected to preserve a contemporaneous record of how the vaccination effort was discussed across national media. The dataset enables researchers to analyze media communication strategies during a nationwide public health emergency, with the broader aim of informing more effective public health messaging through mass media. This project represents a collaborative effort between the Yale School of Medicine and the Tobin Center for Economic Policy. Data and Data Collection Overview This collection comprises 5,383 unique publication links from 20 major news outlets—including television networks, magazines, and newspapers—published between December 1, 2020, and September 30, 2021. Only articles that were freely accessible online without subscription or paywall restrictions were included. Articles were collected by the research team (specifically AM) between August 2021 and November 2023 and in April 2024 (by AM and AG). These 20 news outlets were selected based on a 2020–2021 survey of 511 U.S. adults, which identified the outlets most commonly used to obtain information about the COVID-19 vaccination program. A full list of news outlets, along with their reported usage and perceived trustworthiness, is provided in Sources_Selection.docx. Online publications were identified using Google search with a custom date range in week-long increments (e.g., 12/01/2020–12/07/2020), using the keyword “vaccine” in combination with the link to the respective news outlet’s website. Search results were manually reviewed by AM according to the following inclusion and exclusion criteria. Inclusion criteria: Articles published on the selected U.S. news outlets websites ending in “.com” or “.co” that relate to the COVID-19 vaccination program; Articles from the selected international news outlets that serve both their country of origin and the U.S. audience (e.g., BBC, The Daily Mail). Exclusion criteria: Articles published on the international news outlets websites that exclusively serve their country of origin (e.g., domains ending in .uk, .ca, etc. without .com, .co); Publications from universities, government agencies, or other organizations not affiliated with major U.S. news outlets (e.g., domains ending in .edu, .gov, .org); Videos without accompanying transcripts; Publications without textual content; Articles referencing vaccines unrelated to COVID-19; Non-English language publications. Selection and Organization of Shared Data The full list of publications is provided in the data file named "News_Outlets_Publications_Full_List." Entries are organized by news outlet (one per tab), then by publication year, month, week, and article title within each tab. For each entry, the list includes the article’s original download date by the research team, file format (e.g., PDF), original link to the publication, and a permanent link record. The list was verified by MC, CA, AV, AG, and AM, with final quality control performed by AM. Each article was assigned a unique identifier in the format: "Article Title – News Outlet Name", ensuring that each entry appears only once in the final dataset. Additional documentation includes this Data Narrative, a document explaining the source selection and an administrative README file.

The COVID Tracking Project was a volunteer organization launched from The Atlantic and dedicated to collecting and publishing the data required to understand the COVID-19 outbreak in the United States. Our dataset was in use by national and local news organizations across the United States and by research projects and agencies worldwide.

Every day, we collected data on COVID-19 testing and patient outcomes from all 50 states, 5 territories, and the District of Columbia by visiting official public health websites for those jurisdictions and entering reported values in a spreadsheet. The files in this dataset represent the entirety of our COVID-19 testing and outcomes data collection from March 7, 2020 to March 7, 2021. This dataset includes official values reported by each state on each day of antigen, antibody, and PCR test result totals; the total number of probable and confirmed cases of COVID-19; the number of people currently hospitalized, in intensive care, and on a ventilator; the total number of confirmed and probable COVID-19 deaths; and more.

Methods This dataset was compiled by about 300 volunteers with The COVID Tracking Project from official sources of state-level COVID-19 data such as websites and press conferences. Every day, a team of about a dozen available volunteers visited these official sources and recorded the publicly reported values in a shared Google Sheet, which was used as a data source to publish the full dataset each day between about 5:30pm and 7pm Eastern time. All our data came from state and territory public health authorities or official statements from state officials. We did not automatically scrape data or attempt to offer a live feed. Our data was gathered and double-checked by humans, and we emphasized accuracy and context over speed. Some data was corrected or backfilled from structured data provided by public health authorities. Additional information about our methods can be found in a series of posts at http://covidtracking.com/analysis-updates.

We offer thanks and heartfelt gratitude for the labor and sacrifice of our volunteers. Volunteers on the Data Entry, Data Quality, and Data Infrastructure teams who granted us permission to use their name publicly are listed in VOLUNTEERS.md.

A study in March 2020 showed that around ** percent of surveyed U.S. adults had encountered a lot or at least some news about the coronavirus which seemed completely made up, highlighting the issue of fake news circulation and unreptuable sources seeking to capitalize on the public's need for news and updates in times of crisis. Interestingly, more right-leaning audiences professed to have come across mae up news than Democrats, with just ** percent of Republicans saying that they had seen made up news about the coronavirus, compared to ** percent of news consumers who left-leaning.

The database includes 97 pairs of Chinese/English news headlines from The Global Times Editorial (GTE) and 77 pairs from The Economist Global Business Review (EGBR) that report on the COVID-19 pandemic

This data set aggregates mainstream U.S. media news articles and opinion show transcripts concerning Covid-19 mask-wearing between April 6 and June 8, 2020. Additionally, for several paragraphs of the news articles, it includes crowd-sourced annotation of the statements according to 14 mask-wearing attitude questions taken from Howard 2020's Face Mask Perception Scale (FMPS). Each annotated paragraph thus contains 14 labels (e.g., for "Does the text presented convey the idea that it is difficult to breathe while wearing a face mask?" options include 0 (it is difficult), 1 (it is not difficult), and 2 (does not mention)) with a confidence score ranging from 0-6 for each label. In total, this data set contains 2,361 news articles from eight sources (Daily Kos, Vox, New York Times, Fox, Breitbart, Tucker Carlson, Laura Ingraham, Sean Hannity), including article title, publication date, source, and raw text. Another file of the 8,473 paragraphs contained in all the articles is included with unique paragraph IDs. A separate file of crowd-sourced annotations is also included where labels are given for certain paragraph IDs, and contains 7,559 total annotations across 297 paragraphs and 202 articles. Instructions for how to load the data, as well as filter the annotations for high-quality versions (where there is high confidence or inner-annotator agreement), can be found at https://github.com/ricknabb/media-ideology-coding.

There are mounting concerns about the adverse effects of social media on the public understanding of the COVID-19 pandemic and its potential effects on vaccination coverage. Yet early studies have focused on generic social media use and been based on cross-sectional data limiting any causal inferences. This study is among the first to provide causal support for the speculation that social media news use leads to vaccine hesitancy among US citizens. This two-wave survey study was conducted in the US using Qualtrics online panel-based recruitment. We employ mediation and moderated mediation analyses to test our assumptions. The results suggest that using social media to consume news content can translate into vaccine hesitancy by increasing citizens’ skepticism regarding the efficacy of vaccines. However, these effects are contingent upon the news literacy of users, as the effects on vaccine hesitancy are more substantial among those with lower news literacy. The current study recommends to public policymakers and vaccine communication strategists that any attempt to reduce vaccine hesitancy in society should factor in the adverse effects of social media news use that can increase vaccine safety concerns.