A dataset of media realeases (Twitter, News and Comments, Youtube, Facebook) form related to COVID-19 for open research

Andrzej Jarynowski, Monika Wójta-Kempa, Małgorzata Stochmal, Daniel Płatek, Martyna Mencel, Jan Maciejewski, Alexander Semenov, Vitaly Belik

Interdisciplinary Research Institute, Wroclaw, Poland; System Modeling Group, Institute of Veterinary Epidemiology and Biostatistics, Freie Universität Berlin, ; and many others

EOSC USE Case 03.09.2020

1 / 1 Outline

I Timeline of events in Poland I The use of the traditional and social-content media on the Internet in Poland I infodemics I Data: Twitter, News and Comments, Youtube, Facebook I Analysis of text records I Analysis of networks

2 / 1 Internet media Poland

In this study, by quantitative analysis of “digital traces/footprints” on the Internet (social and traditional media), we concentrate on the number and nature of events such as information queries and information propagation. We cover ca. 28 millions speaking residents of Poland representing over 85% of the literate population. The Polish Internet users are younger (especially among content creators) with females prevailing (especially among content receivers) as compared to the average population. Our results do not describe the perception of the oldest groups, due to digital exclusions, health literacy limitations or consideration of social media as less informative

3 / 1 Interest in “Coronavirus” in Poland

16.03 services 24.03 closure mobility restriction

29.01 31.01 16.04 First case 27.02 04.03 25.05 WHO declares 02-03.03 Restriction In Germany Fake news first case Restriction 16.06 Global threat Special releasing On first case In Poland 25.04 releasing Restriction Act plan 25.01 In Poland New 2 releasing MD Liang exams 3 dates Wudong 11.03 death Pandemic and 31.03 school closure 04.04 Epidemic Anticrisis declaration Anticrisis And law Mitigation Physical Law 2.0 Italian Waiting phase distansing Restrictions releasing Chinese phase phase Phase Anticrisis

We have identified several temporal major clusters of interest on the topic COVID-19: 1) Chinese, 2) Italian, 3) Waiting, 4) Mitigations, 5) Social distancing and Lockdown, 6) Anti-crisis shield, 7) Restrictions releasing, 8) Incease of incidence.

4 / 1 SYTUACJA OGÓLNOPOLSKA Phases

PHASES EMOTIONS TOPICS

Chinese neutral economics 15.01-17.02

Italian negative healthcare system 18-27.02

waiting diverse politics 28.02-09.03

declaration of restrictions fear restrictions 10.03-17.03

Stay at home positive social 18.03-31.03

anticrisis shield diverse epidemiology 01.04-07.05

restrictions releasing optimism politics and economy 08.05-25.07 5 / 1

incidence growth fear social 26.07-05.08

Intensity of the interest 0 25 50 75 100 in the Internet Interest in “Coronavirus” did not reflect transmision dynamics

Moreover, “Are Online Searches for the Novel Coronavirus (COVID-19) Related to Media or Epidemiology? A Cross-sectional Study”, published in International Journal of Infectious Diseases (doi: https://doi.org/10.1016/j.ijid.2020.06.028) presented the case important conclusions. that online searches for COVID-19 in are not correlated with epidemiology.

6 / 1 Interest in “Coronavirus” could reflect transmision dynamics in second wave

Incease of both incidence and interests in last weeks

7 / 1 Infodemic

I Traditional vs social-containment media I a concept in the social medicine area: “lay-referral framing system” where opinions and beliefs of the general public are something different from medical knowledge. I information ws widely treated as a trigger of consequences impacting daily-life security I declarations of mitigation strategies by the prime minister or the minister of health gathered the highest attention to the COVID-19 pandemic of the Internet users in Poland. Thus enacted or in force events do not affect the interest to such extent. However, databases on mitigation strategies (WHO, ECDC, etc.) do not keep this in their records, although it would be of great importance.

8 / 1 The Power of Fake News

RSV by Voivevodship 100 90 First case 80 in Lubusz 70 Fake case Outbreaks 60 in Lodz in 50 V

S 40 R 30

20

10

0 5 9 3 7 1 4 8 2 6 0 4 8 3 7 1 5 9 3 7 1 4 8 2 6 0 4 8 2 6 0 4 8 2 6 0 3 7 1 5 9 3 7 -1 -1 -2 -2 -3 -0 -0 -1 -1 -2 -2 -2 -0 -0 -1 -1 -1 -2 -2 -3 -0 -0 -1 -1 -2 -2 -2 -0 -0 -1 -1 -1 -2 -2 -3 -0 -0 -1 -1 -1 -2 -2 1 1 1 1 1 2 2 2 2 2 2 2 3 3 3 3 3 3 3 3 4 4 4 4 4 4 4 5 5 5 5 5 5 5 5 6 6 6 6 6 6 6 -0 -0 -0 -0 -0 -0 -0 -0 -0 -0 -0 -0 -0 -0 -0 -0 -0 -0 -0 -0 -0 -0 -0 -0 -0 -0 -0 -0 -0 -0 -0 -0 -0 -0 -0 -0 -0 -0 -0 -0 -0 -0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2

koronawirus: (Kuyavian-Pomeranian ) koronawirus: ( Voivodeship) koronawirus: ( Voivodeship)

koronawirus: (lodz Voivodeship) koronawirus: (Lower ) koronawirus: ( Voivodeship)

koronawirus: () koronawirus: () koronawirus: ( Voivodeship)

koronawirus: (Lower Silesian Voivodeship) koronawirus: () koronawirus: (Silesian Voivodeship)

koronawirus: (Subcarpathian Voivodeship) koronawirus: (Swietokrzyskie) koronawirus: (Warmian-Masurian Voivodeship)

koronawirus: (West Pomeranian Voivodeship)

An effect of fake news of the first confirmed case

9 / 1 The Power of Information

RSV by Voivevodship 100 90 80

70 Lubuskie 60 Zachodniopomorskie Warmińsko-Mazurskie 50 Dolnośląske

V 40 S Śląskie R 30 Mazowieckie Małopolskie 20 Wielkopolskie 10 Lubelskie day Opolskie 0 Podkarpackie Łódzkie Świętokrzyskie Pomorskie koronawirus: (Kuyavian-Pomeranian Voivodeship) koronawirus: (Greater Poland Voivodeship) Podlaskie Kujawsko-Pomorskie koronawirus: (Lesser Poland Voivodeship) koronawirus: (lodz Voivodeship) 0 6 2 8 koronawirus: (Lower Silesian Voivodeship) koronawirus: () 1 1 koronawirus: (Lubusz Voivodeship) koronawirus: (Masovian Voivodeship) koronawirus: () koronawirus: (Lower Silesian Voivodeship) koronawirus: (Pomeranian Voivodeship) koronawirus: (Silesian Voivodeship) koronawirus: (Subcarpathian Voivodeship) koronawirus: (Swietokrzyskie) koronawirus: (Warmian-Masurian Voivodeship) koronawirus: (West Pomeranian Voivodeship)

Introduction day as the main driving force

10 / 1 Datasets (deposited)

https://zenodo.org/record/3985807 I extracted 57,306 representative articles in Polish using Eventregitry.org tool in language Polish and topic “Coronavirus” in article body; I extracted 1,015,199 and Tweets from #Koronawirus in language Polish usiing Twitter API. I collected 1,574 videos with keyword: Koronawirus on YouTube and 247,575 comments on them using Google API;

11 / 1 Datasets (deposited) 2

I manually labeled 1,449 articles / Facebook posts from and 111 texts from outside this region; I manually labeled 1,500 Twitts / Youtube comments (for ML training); I analysis of 244 social empirical studies till 25.05 on COVID-19 in Poland.

12 / 1 Techniques

Methodological and statistical analysis is a crucial point in this kind of study. We apply: I the number and nature of social media events such as information queries (time series analysis); I sentiment and conceptual fields analysis; I social network analysis; I topic modeling technique; I comparative approach bewtween local and national-wide media; I quasi-systematic review of studies.

13 / 1 Twitter

I We have collected >1 000 000 with #Koronawiurs I Twitter in Poland has relatively low popularity (~3 million registered users or less than 8% of the country population) and is mainly used by expats, journalists and politicians I For Twitter data we will deploy the temporal multilayer network analysis of users (using metadata on quotes, retweets, mentions, replies, and followings), because networks can be constructed based on the mentioned metadata types.

14 / 1 Retweet networks

●

● ● ●

● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ●● ● ● ● ● M_K_Blonska●●● ● ● ● ●●● ● ●● ● ● ● ● ●MichalSzczerba● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●

● ●

● ● ● ● ●

● ● ● ● ● ● ● ● ● ● ● ● ●● ● AndrzejTurczyn● ●●● ● ●● ● ●● ● ● ● ● ● ● ● ● ●● MZ_GOV_PL● ● ● ● ● ● ● ● ● ● ● ●● ● ●● ● ● ● ● ● ● ● PremierRP● ● ● ● ●● ● ●●● ● ● ● ● ● ● ● ● ●● ● ● ● ● KONFEDERACJA_ ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●

● ● ● ● ● ● ● ● ● ● ● ● ●

●

● ●

●

● ● A network of retweets with the Koronawirus hashtag in Polish language ● during 28.01-11.03.2020 period (250 most central nodes according to

weighted centrality, coloring according to● Louvain community detection algorithm). Yellow – the ruling coalition, green – the mainstream opposition, dark blue – a Protestant minority, light blue – a far right party ●

15 / 1 Polarization

0.9 net_layer 0.8 mod_retweet mod_reply mod_quote modularity 0.7

0.6

10 20 30 weeks

Modularity of the full networks (over 30 thousands accounts) and their weekly dynamics for different type of communication relations (retweeting, reply, quotation).

16 / 1 Actors

The most central users

17 / 1 News articles

I We have collected >60000 epresentative digital traditional media news (selected with EventRegistry tool) with stemmed word Koronawirus in title or short body. I We choose EventRegistry (eventregistry.org) as a traditional media search engine because it has a large range of online news services representing a broad political spectrum. Besides, it gives priority to digital versions of other broadcasting channels, including television, radio or newspapers. I For newss data we will deploy the temporal multilayer network analysis of aricles or news agency based on context similarity

18 / 1 Article sources

The most frequent sources

19 / 1 Article sources time dynamics

20 / 1

The most frequent sources frequency of sources changing with time Youtube

Good source for cultural effect of COVID and monitoring protest movements 21 / 1 Sociolingistics

New Dictionary for Corpus

Intrest in masks

22 / 1 Topic Modelling

Weekly dynamics of media perception in Poland. Main Topics in over 50 thousand news articles (without garbage codes).

23 / 1 Topics

Topics manually coded

24 / 1 Sentiment

Sentiment across ML labeled texts

Sentiment across manually labeled texts 25 / 1 Empirical Social Studies

I 180 projects carried out by scientific and commercial institutions in the period from January to May 2020. I present a standard way of conducting empirical studies by social researchers who undertake the challenge of recognizing pandemic conditions and challenges in the future I mitigating the polarization of social attitudes which are dynamically changing during the pandemic, I practical solving / and not only diagnosing / problems revealed in covid reality I supplementing the deficiencies of theoretical assumptions accompanying research projects.

26 / 1 Empirical Social Studies factor analysis

Main factors of social studies objects

27 / 1 Intervention schema

Risk Seeking

Risk Reduction

Risk Seeking

Risk Reduction

An attempt to describe main aspects of a Polish social system in Pandemic time. Rectangles are subpopulations, arrows mean main interactions, shadow variable are modifiable factors, signs represents positive or negative influence, crossing bars means interaction which are the most likely to disappear with time.

28 / 1 Conlusions

I The COVID-19 epidemic in Poland is primarily of a social, not medical, dimension; I Local media are more reliably than national media, and Lower seem to be less susceptible to media manipulation and conspiracy theories; I Campaigns should be targeted to sccific population and regions; I Declarations of response strategies by the Polish prime minister or the minister of health gathered the highest attention of Internet users. So already enacted or in force events do not affect the interest to such an extent; I The comments are significantly more likely to be negative. www.infodemia-koronawirusa.pl

29 / 1