trigram

This is a table of type trigram and their frequencies. Use it to search & browse the list to learn more about your study carrel.

trigram	frequency
the number of	269
as well as	144
in order to	129
on social media	122
of social media	117
one of the	105
based on the	105
in terms of	89
the united states	87
number of tweets	84
more likely to	82
due to the	81
in social media	79
the use of	78
the spread of	69
a number of	66
in this study	62
during the covid	61
in the united	58
of altmetric data	56
of the tweets	56
the impact of	54
of the pandemic	53
likelihood of sharing	53
such as the	52
analysis of the	51
likely to be	51
social media data	50
the context of	50
the case of	48
social media platforms	47
hot research topics	47
social media and	46
a set of	45
to understand the	45
of the most	44
the fact that	44
higher levels of	43
the presence of	43
total number of	42
as a result	42
can be used	41
there is a	41
in this paper	41
of the covid	39
natural language processing	39
related to the	39
the social media	39
according to the	38
as shown in	38
be used to	37
most of the	36
in the context	35
the relationship between	34
the majority of	34
well as the	34
local tv viewership	34
each of the	34
at the time	34
the distribution of	33
of this study	33
on the other	33
of the virus	33
the time of	32
in addition to	32
a list of	32
we used the	32
the total number	31
an analysis of	31
in the same	31
the role of	31
the results of	30
during the pandemic	30
it is possible	30
used in the	29
is based on	29
in this section	29
the effect of	29
in the future	29
shown in fig	29
world health organization	28
we found that	28
to display the	28
the importance of	28
a total of	28
associated with the	28
of the tweet	28
in the first	28
stages of the	28
in the uk	28
and social media	28
granted medrxiv a	27
who has granted	27
the percentage of	27
the age of	27
has granted medrxiv	27
medrxiv a license	27
the other hand	27
some of the	27
of tweets in	27
license to display	27
the field of	27
of machine learning	27
preprint in perpetuity	27
use of twitter	27
a license to	27
the preprint in	27
display the preprint	27
of the outbreak	27
all of the	26
to analyze the	26
international license it	25
available under a	25
the frequency of	25
of twitter users	25
to identify the	25
it is made	25
in this case	25
made available under	25
the analysis of	25
the copyright holder	25
is made available	25
and machine learning	25
analysis of tweets	25
license it is	25
analysis of twitter	25
social big data	24
are more likely	24
holder for this	24
it is a	24
of twitter followers	24
were more likely	24
spread of the	24
such as twitter	24
the rest of	24
table shows the	24
use of social	24
the performance of	24
for this preprint	24
social media users	24
of hate speech	24
copyright holder for	24
in the field	23
in which the	23
social media use	23
is the author	23
team financial performance	23
state of alarm	23
to be a	23
the author funder	23
of the data	23
shown in table	23
social media to	23
the end of	23
there is no	23
in the case	23
the beginning of	22
was used to	22
from social media	22
tweets in the	22
and the number	22
the purpose of	22
of brand equity	22
found that the	22
the same time	21
in the age	21
it is also	21
content analysis of	21
of the public	21
the set of	21
because of the	21
it can be	21
were used to	21
use of the	21
under a is	21
words in the	21
of tweets and	21
a is the	21
the proportion of	21
compared to the	21
are shown in	21
shown in figure	21
to the covid	21
figure shows the	21
this preprint this	20
are related to	20
the probability of	20
in the data	20
the content of	20
social media is	20
preprint this version	20
using social media	20
this version posted	20
and local tv	20
of the crisis	20
around the world	20
the size of	20
of the number	20
the amount of	20
of the two	20
a case study	20
major league baseball	19
to the pandemic	19
the h n	19
a variety of	19
the most popular	19
in the following	19
rest of the	19
use social media	19
for public health	19
of the network	19
is consistent with	19
of twitter data	19
this is the	19
we observe that	19
international conference on	19
can be seen	19
in this work	19
it is worth	19
the most common	19
of the users	19
number of retweets	19
we see that	18
in the social	18
are likely to	18
of all the	18
is important to	18
it is not	18
related to covid	18
of the three	18
of tweets that	18
part of the	18
can also be	18
in the number	18
at least one	18
to the public	18
x x x	18
latent dirichlet allocation	18
is one of	18
measure of concern	18
smt immunity link	18
number of publications	18
based on a	18
social network analysis	18
a result of	18
the need for	18
distribution of the	18
response to the	18
with the same	18
relationship between non	17
a smt immunity	17
to predict the	17
social media in	17
included in the	17
the effectiveness of	17
using twitter data	17
it has been	17
the tweets in	17
the behavior of	17
in response to	17
we find that	17
focus on the	17
it would be	17
of verified users	17
in the us	17
the degree of	17
to focus on	17
in support of	17
a sample of	17
in the past	17
that have been	17
and so on	17
it is important	17
is the most	16
number of users	16
large number of	16
is that the	16
in real time	16
which is the	16
of sharing the	16
is used to	16
that can be	16
a combination of	16
words related to	16
the problem of	16
that there is	16
high level of	16
of the social	16
the level of	16
to detect the	16
the coverage of	16
words such as	16
shows that the	16
tweets related to	16
can be found	16
for each user	16
the twitter api	16
an overview of	16
of the main	16
is associated with	16
in the tweets	16
the most important	16
the public health	16
with respect to	16
members of the	16
are able to	16
in the present	16
appears to be	16
early stages of	15
is possible that	15
the usage of	15
in relation to	15
it should be	15
impact of covid	15
at the same	15
for social media	15
high levels of	15
of words in	15
for each of	15
on twitter during	15
a social media	15
focused on the	15
to do this	15
perception of the	15
social media for	15
with regard to	15
terms of the	15
levels of abuse	15
the measure of	15
new media literacy	15
are used to	15
number of replies	15
what are the	15
of tweets during	15
it is the	15
time of the	15
the evolution of	15
which is a	15
the name of	15
the effects of	15
the machine learning	15
coverage and intensity	15
that it is	15
are based on	15
is an important	15
majority of the	15
in the form	15
used as a	15
number of times	15
that machine learning	15
in times of	15
the form of	15
likely to share	14
in the network	14
version posted may	14
a subset of	14
democrat and republican	14
case of the	14
a comparison of	14
number of posts	14
resilient communication ecosystem	14
derived from the	14
machine learning algorithms	14
twitter as a	14
found in the	14
increase in the	14
the early stages	14
nature of the	14
we use the	14
is possible to	14
the likelihood of	14
and it is	14
used in this	14
theory of brand	14
different types of	14
tweets that are	14
to be the	14
this study is	14
may not be	14
in the usa	14
a series of	14
the extent to	14
unified theory of	14
extent to which	14
of the news	14
the world health	14
the correlation between	14
for the period	14
this is a	14
is shown in	14
support of h	14
tweets that were	14
from march to	14
tweets during the	14
in social networks	14
in other words	14
to capture the	14
associated with a	14
needs to be	14
an example of	14
gender differences in	14
of fake news	14
the data collection	14
of all abuse	14
the concept of	14
personal versus news	14
the new york	14
during the first	14
of social distancing	14
the basis of	14
users in the	14
understanding of the	14
this means that	13
social distancing and	13
many of the	13
from the twitter	13
in the twitter	13
depending on the	13
the novel coronavirus	13
the rise of	13
the validated network	13
we use a	13
are associated with	13
the peaks of	13
semantic network analysis	13
consistent with the	13
based on their	13
using the twitter	13
and number of	13
is defined as	13
season marketing assets	13
a personal tweet	13
lower levels of	13
from january to	13
social media as	13
tweets and the	13
relationship between in	13
how social media	13
on twitter and	13
reported likelihood of	13
the goal of	13
of the information	13
for each tweet	13
the state of	13
the tweet is	13
we propose a	13
in the early	13
levels of twitter	13
of the word	13
for all the	13
likely to have	13
the influence of	13
for this reason	13
the definition of	13
number of topics	13
in the previous	13
the label propagation	13
described in section	13
in the world	13
were associated with	13
the potential to	13
the volume of	13
different stages of	13
in this context	13
the maintenance stage	13
to extract the	13
size of the	13
with higher levels	13
marketing assets and	13
in the analysis	13
during the early	13
the list of	13
personal negative tweets	13
in our dataset	13
as opposed to	13
of the total	13
of the first	13
h n outbreak	13
and found that	13
negative versus non	12
a high level	12
on the basis	12
of public health	12
times of crisis	12
the measles outbreak	12
the higher the	12
network of verified	12
results of the	12
online social networks	12
would like to	12
supported by the	12
we want to	12
of the paper	12
be able to	12
it does not	12
publications in the	12
state of the	12
seasonal marketing assets	12
the geocov tweets	12
other social media	12
was associated with	12
the dynamics of	12
democrats and republicans	12
be seen in	12
right and right	12
in our case	12
as it is	12
to determine the	12
the tweets were	12
information about the	12
be used as	12
been shown to	12
can be applied	12
each hashtag group	12
and right wing	12
number of followers	12
state sponsored media	12
performance of the	12
of the political	12
a review of	12
refers to the	12
geocov tweets dataset	12
during this time	12
in the literature	12
all the tweets	12
in patients with	12
the frames used	12
in the dataset	12
attendance and local	12
a measure of	12
the tweets and	12
a lot of	12
is able to	12
is necessary to	12
features of the	12
with the help	12
the formation of	12
found to be	12
represented by the	12
defined as the	12
positive and negative	12
the nature of	12
the perception of	12
during the h	12
machine learning systems	12
appear to be	12
with a high	12
the value of	12
on social networks	12
the start of	12
online social media	12
the help of	12
topics in the	12
public health frame	12
it is necessary	11
during this period	11
media use in	11
in this chapter	11
need to be	11
we created a	11
on the same	11
characteristics of the	11
twitter has been	11
with the highest	11
more than one	11
to the fact	11
the existence of	11
we used a	11
the pandemic and	11
the appearance of	11
seen in figure	11
in the time	11
social media during	11
the possibility of	11
most altmetric data	11
of the current	11
context of the	11
presence of altmetric	11
be noted that	11
of the disease	11
network analysis of	11
a large number	11
a tweet is	11
which can be	11
the university of	11
highest number of	11
per number of	11
in the digital	11
new york times	11
changes in the	11
we can find	11
of the study	11
that social media	11
sentiment analysis and	11
versus news classification	11
text of the	11
online hate speech	11
media coverage of	11
diffusion patterns of	11
there is an	11
out of the	11
coverage of the	11
number of population	11
as one of	11
diversion and suppression	11
the general public	11
relationship between the	11
official accounts of	11
to measure the	11
to account for	11
levels of attention	11
and in the	11
be found in	11
the pandemic in	11
deaths per number	11
of conspiracy narratives	11
in regard to	11
the lack of	11
refer to the	11
resulting in a	11
across subject fields	11
the topic of	11
the absence of	11
each of these	11
two types of	11
platforms such as	11
a function of	11
in this article	11
to examine the	11
social networking sites	11
in the appendix	11
as the pandemic	11
types of altmetric	11
and team financial	11
the twitter data	11
social media has	11
machine learning is	11
frustration and hope	11
more than million	11
of the population	11
the date of	11
to understand how	11
this kind of	11
a health crisis	11
for the covid	11
of the world	11
the types of	11
the highest number	11
showed that the	11
words associated with	11
results show that	11
this data is	11
representation of the	11
higher than the	11
age of twitter	11
the first step	11
along with the	11
purpose of this	11
is likely to	10
this research is	10
labeled as a	10
to study the	10
hate speech detection	10
can be a	10
the structure of	10
the most frequently	10
by twitter users	10
machine learning and	10
allows us to	10
may lead to	10
to have a	10
that were not	10
through social media	10
have been used	10
the tweets that	10
related to a	10
is the case	10
of abusive replies	10
contribute to the	10
the network of	10
a survey of	10
as a tool	10
that epic m	10
in the last	10
to find the	10
known to be	10
we need to	10
social media platform	10
defined as a	10
of this research	10
validated network of	10
a resilient communication	10
digital media literacy	10
that were used	10
pandemics in the	10
set of tweets	10
result of the	10
facebook and instagram	10
different from the	10
of all tweets	10
caused by the	10
of the twitter	10
the social network	10
daily number of	10
as discussed in	10
be more likely	10
we also observe	10
in a tweet	10
the tweet text	10
in the tweet	10
were able to	10
using the same	10
web of science	10
can be defined	10
be applied to	10
we can see	10
there are two	10
in the use	10
that most of	10
hair et al	10
of the model	10
and can be	10
and public health	10
be defined as	10
in recent years	10
in a given	10
between the two	10
was based on	10
as a function	10
in the supplementary	10
to reduce the	10
the unified theory	10
the application of	10
the world and	10
public health professional	10
such as facebook	10
political orientation of	10
for the purpose	10
during a crisis	10
via social media	10
fuzzy accuracy assessment	10
can lead to	10
is not a	10
as the number	10
role in the	10
language processing and	10
tweets containing the	10
version of the	10
to the best	10
that they had	10
frames used in	10
social media can	10
based on twitter	10
we present the	10
a social network	10
to explore the	10
to deal with	10
the fields of	10
in the u	10
level of attention	10
to the data	10
the following tweet	10
rather than a	10
of social networks	10
should be noted	10
in online social	10
the recovery period	10
altmetric data for	10
portion of the	10
content of the	10
support vector machine	10
a news tweet	10
an increase in	10
case study of	10
that the model	10
the data from	10
the part of	10
a semantic network	9
for disease control	9
in line with	9
to social media	9
of a tweet	9
and game attendance	9
such as a	9
social and machine	9
research topics are	9
it may be	9
policy document citations	9
description of the	9
proceedings of the	9
while it is	9
the middle of	9
i is the	9
data presence across	9
the collected data	9
are the most	9
was used as	9
us presidential election	9
the diffusion patterns	9
that we are	9
support for the	9
in our sample	9
the first two	9
to provide a	9
machine learning methods	9
the vast majority	9
summarised in table	9
level of abuse	9
was used in	9
higher likelihood of	9
on the topic	9
by the public	9
the words that	9
news update frame	9
followed by the	9
the collection of	9
extracted from the	9
at this time	9
that our model	9
the democrat and	9
cov tweets dataset	9
to social distancing	9
were likely to	9
we set the	9
as a personal	9
the emergence of	9
seen them before	9
in conjunction with	9
this type of	9
for this purpose	9
research topics in	9
we are able	9
in the current	9
event detection from	9
to changes in	9
the best of	9
vast majority of	9
a public health	9
the vaccine frame	9
of the topics	9
relevant to the	9
for event detection	9
on twitter data	9
relative to the	9
the focus of	9
to be retweeted	9
that there are	9
for each word	9
has been used	9
as a percentage	9
of uk mps	9
number of covid	9
the path between	9
for sentiment analysis	9
of the existing	9
the political frame	9
which was not	9
marketing assets are	9
to investigate the	9
abuse sent to	9
twitter data to	9
to the number	9
on a large	9
positively related to	9
the result of	9
is labeled as	9
percentage of the	9
this is consistent	9
review of the	9
words that are	9
number of cases	9
note that the	9
tweets posted by	9
of social and	9
information of the	9
fact that the	9
to work with	9
in the next	9
the notion of	9
we did not	9
it is likely	9
shown in the	9
worth noting that	9
is represented by	9
shows the results	9
all of these	9
the development of	9
media and the	9
related to social	9
the findings of	9
cases and deaths	9
the twitter users	9
contribution to the	9
structure of the	9
has also been	9
similar to the	9
the onset of	9
perception of risk	9
number of deaths	9
have shown that	9
topics that are	9
of the user	9
twitter rest api	9
sales and marketing	9
the popularity of	9
in the second	9
order to identify	9
sent to mps	9
the period of	9
with the number	9
is designed to	9
overview of the	9
at the university	9
on the relationship	9
a sentiment analysis	9
on twitter in	9
between in season	9
the word co	9
on the internet	9
the potential of	9
the daily number	9
women of colour	9
the word cloud	9
in figure we	9
and social distancing	9
the scope of	9
more than of	9
table presents the	9
evolution of the	9
abuse for the	9
the pandemic has	9
the political orientation	9
seasonal ma and	9
of data collection	9
as hot research	9
of this paper	9
of coverage and	9
data from social	9
of abuse for	9
right wing parties	9
tend to be	9
by the number	9
by the authors	9
twitter during the	9
for this study	9
all abuse sent	9
on boris johnson	9
is worth noting	9
context of covid	9
can be represented	9
social media messages	8
the sir model	8
sample of tweets	8
allows users to	8
we consider the	8
number of mentions	8
of likelihood of	8
on the data	8
altmetric data with	8
while there are	8
social media activities	8
the idea that	8
we aim to	8
do not have	8
the official accounts	8
that twitter users	8
the time series	8
present in the	8
and team financials	8
are presented in	8
greater likelihood of	8
of million tweets	8
parts of the	8
allow users to	8
o o f	8
detecting influenza epidemics	8
the actions of	8
the tweet was	8
the results are	8
on the part	8
tweets in our	8
in the death	8
peer review comments	8
the three disinformation	8
and el mundo	8
tweets that contain	8
conflict of interest	8
the public to	8
the mention network	8
of this article	8
the risk of	8
of the topic	8
the supplementary information	8
disease control and	8
information on social	8
big data and	8
data from twitter	8
of the new	8
if it is	8
to machine learning	8
the action words	8
was significantly predicted	8
p r o	8
activity of the	8
symptoms of covid	8
there are several	8
in the study	8
in each country	8
less likely to	8
goal of this	8
l p r	8
to compare the	8
into account the	8
but also to	8
to use the	8
the average of	8
data collection and	8
discussion of the	8
show that the	8
the analysis was	8
most prominent news	8
to calculate the	8
to note that	8
corresponding to the	8
the sentiment analysis	8
addition to the	8
be useful to	8
are interested in	8
on the social	8
the us presidential	8
the potential reach	8
in studies and	8
to obtain the	8
the news frames	8
the power of	8
the last years	8
identified in the	8
the quality of	8
j o u	8
data mining techniques	8
see appendix a	8
the covid crisis	8
of replies that	8
if the tweet	8
across social media	8
whether or not	8
use of hashtags	8
data from the	8
to be true	8
in the covid	8
compared to other	8
be explained by	8
personal protective equipment	8
the authors declare	8
of twitter use	8
be due to	8
goal is to	8
we are interested	8
to the president	8
were asked to	8
significantly predicted by	8
a l p	8
positive or negative	8
united states during	8
tweets can be	8
is that it	8
the most frequent	8
during the period	8
the initial stage	8
in each tweet	8
a tool for	8
deaths per one	8
specific altmetric data	8
to ensure that	8
in this category	8
we see a	8
shared untrue material	8
in the corpus	8
figure illustrates the	8
to mps in	8
cascades on twitter	8
there are also	8
static and dynamic	8
to get the	8
the united kingdom	8
the expanded analysis	8
altmetric data presence	8
the steel blue	8
the text of	8
the characteristics of	8
topics of interest	8
at the end	8
of the collected	8
mlb brand equity	8
peaks of news	8
u r n	8
and the other	8
tweets and retweets	8
that do not	8
across the world	8
the data is	8
source of information	8
to some extent	8
to the right	8
in our model	8
united states and	8
is presented in	8
words in each	8
of psychophysical numbing	8
as in study	8
it difficult to	8
the same way	8
and data analysis	8
and the public	8
as part of	8
a systematic review	8
we were able	8
gorrell et al	8
of our knowledge	8
observe that the	8
the identification of	8
this is not	8
means that the	8
in which they	8
the fraction of	8
of the analysis	8
on the number	8
to create a	8
the european union	8
altmetric data sources	8
a greater likelihood	8
user u i	8
the accuracy of	8
social media sites	8
on the analysis	8
the activity of	8
and there are	8
pandemic in the	8
is required to	8
about the virus	8
in our analysis	8
p r e	8
this paper is	8
to make the	8
on the web	8
a study of	8
properties of the	8
to monitor the	8
the supplementary material	8
development of a	8
the confusion matrix	8
of alarm and	8
r n a	8
using twitter to	8
used to detect	8
chen et al	8
a sense of	8
in the center	8
phrase i in	8
r o o	8
the accounts of	8
the data and	8
the authors also	8
they had seen	8
of attention on	8
days of the	8
users who reported	8
aspects of the	8
also be used	8
is also a	8
with each other	8
we plan to	8
are consistent with	8
the kinds of	8
has led to	8
correlation between the	8
by the user	8
n a l	8
is an open	8
media such as	8
data in the	8
number of confirmed	8
to estimate the	8
public response to	8
there was a	8
twitter data for	8
social media are	8
of users and	8
and spread of	8
with at least	8
and how they	8
been applied to	8
directed validated network	8
the raw data	8
have the potential	8
the process of	8
and domestic abuse	8
indicated that the	8
the ability to	8
o u r	8
that of the	8
some of these	8
impact of the	7
the need to	7
that the public	7
how the public	7
beginning of the	7
a percentage of	7
words in a	7
political frame was	7
is provided in	7
the frequency with	7
take into account	7
the burden on	7
used to identify	7
the second step	7
is the number	7
or of the	7
the stories were	7
topics related to	7
sciences and humanities	7
that are related	7
to try to	7
twitter streaming api	7
a crucial role	7
as measured by	7
assets and team	7
may be more	7
and th inclusive	7
by using the	7
were used for	7
of the problem	7
orientation of users	7
is the first	7
that are currently	7
q a mentions	7
increasing number of	7
are provided in	7
tweets concerned with	7
between non seasonal	7
inverse document frequency	7
the same as	7
the bipartite network	7
they may be	7
april st and	7
non seasonal and	7
research topics with	7
data for the	7
on day t	7
social sciences and	7
to jurisdictional claims	7
any of the	7
the time period	7
account for the	7
to better understand	7
twitter sentiment analysis	7
the first three	7
mps between april	7
to include the	7
was found to	7
during times of	7
and risk communication	7
be related to	7
the positive relationship	7
and after the	7
to get a	7
there have been	7
for this task	7
note springer nature	7
march to april	7
of data mining	7
the dependent variable	7
of the event	7
the directed validated	7
we will use	7
percentage of replies	7
historical sharing of	7
terms such as	7
context in which	7
information cascades on	7
social distancing rules	7
social media mining	7
pew research center	7
nodes in the	7
is a significant	7
regard to jurisdictional	7
to mps between	7
public perception of	7
to represent the	7
tweets collected between	7
claims in published	7
and likelihood of	7
for the first	7
as a predictor	7
associated with higher	7
to reflect the	7
so as to	7
in season and	7
on facebook and	7
in each hashtag	7
mcrobbie and thornton	7
a lack of	7
tweets in each	7
topic modeling and	7
accounts in the	7
in more detail	7
the top ten	7
the data for	7
users on twitter	7
in the validated	7
to obtain a	7
is organized as	7
potential reach of	7
that the results	7
to each other	7
terms related to	7
at the micro	7
to see if	7
there is evidence	7
machine learning framework	7
fake news detection	7
the complexity of	7
amount of data	7
the data were	7
of the art	7
we will discuss	7
is related to	7
i in t	7
the study was	7
the one of	7
frequency with which	7
the seriousness of	7
the work of	7
the increase in	7
to increase the	7
subject fields and	7
positive relationship between	7
to visualize the	7
jurisdictional claims in	7
in case of	7
by public health	7
number of the	7
severe acute respiratory	7
on how to	7
exploratory analysis of	7
correlated with the	7
to a specific	7
shown to be	7
look at the	7
name of the	7
to support the	7
of the models	7
of the presence	7
refuting a smt	7
end of the	7
and with the	7
used for the	7
we calculate the	7
twitter users are	7
tweets using the	7
we chose to	7
stress symptoms and	7
of the content	7
areas such as	7
has been a	7
the early days	7
behavior of users	7
of a user	7
media as a	7
is a common	7
have been widely	7
protection motivation theory	7
aim of this	7
the fuzzy accuracy	7
neutral with regard	7
people are more	7
between april st	7
predicted by higher	7
the same day	7
the corexq algorithm	7
we believe that	7
nature remains neutral	7
axis represents the	7
the news update	7
used to extract	7
if a tweet	7
to assess the	7
a multiple regression	7
number of words	7
in published maps	7
the likelihood that	7
using natural language	7
of information and	7
study is to	7
this is also	7
to be more	7
public health concerns	7
public health crisis	7
used in study	7
the brand equity	7
allowed us to	7
the era of	7
the task of	7
the sampling period	7
the mueller investigation	7
the tweets are	7
to the twitter	7
it is interesting	7
to the information	7
due to their	7
that the tweet	7
to answer the	7
altmetric data across	7
seem to be	7
reported by the	7
in each group	7
there are a	7
promoting or refuting	7
likelihood of having	7
of this work	7
significantly from to	7
has been widely	7
early days of	7
altmetric data in	7
a machine learning	7
was a significant	7
tweets about covid	7
the mental health	7
of the italian	7
study of the	7
differences in the	7
the last two	7
do not necessarily	7
period of time	7
we have the	7
twitter event detection	7
the study period	7
we focus on	7
of the epidemic	7
community structure in	7
is a need	7
been found to	7
is difficult to	7
patterns of covid	7
the data in	7
media data for	7
negative matrix factorization	7
divided by the	7
different altmetric data	7
fields of science	7
based brand equity	7
the introduction of	7
big data technologies	7
the authors have	7
research on the	7
such as data	7
crucial role in	7
we also see	7
the mapreduce paradigm	7
the course of	7
and institutional affiliations	7
the natural language	7
for our analysis	7
focuses on the	7
same way as	7
appeared to be	7
to one of	7
over million tweets	7
to identify and	7
is worth mentioning	7
most frequently used	7
of words associated	7
st and th	7
to extend the	7
during the last	7
to engage with	7
have also been	7
as an example	7
sentiment classification method	7
context of a	7
social science research	7
with machine learning	7
more and more	7
wang et al	7
the twitter rest	7
maps and institutional	7
and reply tweets	7
to improve the	7
the sharing of	7
of twitter in	7
during the maintenance	7
the aim of	7
is available at	7
is not the	7
published maps and	7
the retweet network	7
social media user	7
this may be	7
and network analysis	7
a framework for	7
start of the	7
the virus and	7
centers for disease	7
been used to	7
the relationships between	7
provided in table	7
social media posts	7
spread of information	7
the order of	7
to the virus	7
for the top	7
social media the	7
personal negative tweet	7
information such as	7
and the news	7
of the united	7
the next step	7
the outbreak of	7
to evaluate the	7
public health events	7
of the impact	7
communities in the	7
was used for	7
in the fields	7
public coronavirus twitter	7
of having seen	7
in natural language	7
onset of the	7
social and economic	7
to express their	7
the public and	7
of the lockdown	7
the diffusion of	7
for the most	7
also known as	7
used to predict	7
time of writing	7
the m s	7
this is an	7
sentiment analysis of	7
the distributions of	7
studies have been	7
for altmetric data	7
be that the	7
springer nature remains	7
is not clear	7
remains neutral with	7
hate speech in	7
effect of the	7
from the data	7
the population to	7
best of our	7
organized as follows	7
all over the	7
intensity of the	7
according to a	7
of each user	7
by using a	7
the output of	7
even in the	7
during a health	7
three disinformation items	7
tweets from the	7
of hot research	7
the chain of	7
the cov tweets	7
the dissemination of	7
to analyse the	7
data mining and	7
the subject of	7
topics of discussion	7
conspiracy narratives on	7
the limitations of	6
before and after	6
in comparison to	6
during the measles	6
a surge in	6
seems to be	6
story attributed to	6
indicators of the	6
we collected the	6
complexity of the	6
in the online	6
analysis of online	6
any kind of	6
of the communities	6
due to its	6
and opinion mining	6
the perspective of	6
in regards to	6
representative of the	6
political news story	6
a focus on	6
we provide a	6
of tweets posted	6
that the most	6
socialdistancing and stayathome	6
to be less	6
themselves as more	6
social media infodemic	6
the advantage of	6
date of the	6
taking into account	6
advances in social	6
social media discourse	6
in the public	6
in table and	6
as a whole	6
an attempt to	6
be interested in	6
panic buying and	6
politicians on twitter	6
disaster risk reduction	6
and covid information	6
based on social	6
the design of	6
march and april	6
with a strong	6
that we have	6
in the text	6
in large networks	6
information ecology framework	6
of disease outbreaks	6
topics on twitter	6
people who are	6
which has been	6
to be untrue	6
during the recovery	6
cases of covid	6
we have a	6
fernandez et al	6
to indicate the	6
uk mps during	6
the top hashtags	6
in the information	6
of the following	6
the sum of	6
piece of information	6
the first time	6
function of the	6
a global pandemic	6
table table table	6
none of the	6
the most dominant	6
outbreak in china	6
influenza epidemics using	6
the management of	6
brand equity model	6
around the coronavirus	6
unigrams and bigrams	6
use of words	6
remains to be	6
number of increased	6
set of words	6
hybrid deep learning	6
the titles of	6
with the stories	6
to see that	6
has been observed	6
kinds of altmetric	6
as a new	6
a tweet that	6
number of other	6
the sensitivity of	6
the combination of	6
we group the	6
of the different	6
machine learning models	6
on top of	6
have been proposed	6
progression of the	6
the details of	6
as of july	6
the panic buying	6
of the pipeline	6
order of magnitude	6
media activities of	6
we compare the	6
and the network	6
a range of	6
the altmetric data	6
results in the	6
users who have	6
working from home	6
on twitter date	6
as an early	6
early detection of	6
the proposed algorithm	6
the extraction of	6
rated likelihood of	6
the news timeline	6
to refer to	6
infected by the	6
the most relevant	6
application programming interface	6
with social media	6
media briefing on	6
word in the	6
applied to the	6
those concerned with	6
as more likely	6
the lens of	6
the sentiment of	6
of the stimulus	6
topics such as	6
the degree sequence	6
is needed to	6
staying at home	6
within the same	6
on the twitter	6
for depression detection	6
towards uk politicians	6
information can be	6
of interest to	6
detection on social	6
version posted june	6
the main topics	6
of the time	6
note that this	6
have been applied	6
a framework to	6
detecting and tracking	6
at the media	6
a large scale	6
ways in which	6
during different stages	6
emerging infectious disease	6
pandemic and the	6
by the virus	6
used as the	6
use of a	6
to conspiracy narratives	6
impact on the	6
time of this	6
first week of	6
the politics of	6
group the hashtags	6
the hashtags into	6
twitter is a	6
on the use	6
ticket sales and	6
the duration of	6
the conservative party	6
replies to mps	6
in a more	6
the lda model	6
words from the	6
large amount of	6
the most active	6
h n pandemic	6
dynamics of hate	6
concerned with fraud	6
opening remarks at	6
subject fields of	6
and firm value	6
twitter mentions and	6
the coronavirus disease	6
to have been	6
we collect data	6
take care of	6
and data mining	6
of twitter to	6
promoting and refuting	6
the initial tweet	6
and news media	6
spread of misinformation	6
machine learning techniques	6
hashtags into six	6
if there is	6
and how it	6
has been applied	6
used by public	6
are a number	6
of tweets was	6
and other social	6
discussed in chapter	6
step sentiment classification	6
game attendance and	6
for those who	6
the next section	6
an eid outbreak	6
in these countries	6
on the left	6
there are some	6
view of the	6
core of the	6
results showed that	6
so that we	6
appear in the	6
that the stories	6
be used for	6
available in the	6
to detect and	6
we note that	6
on twitter is	6
in the era	6
there are multiple	6
hate speech is	6
we present a	6
make sense of	6
it has also	6
play a crucial	6
the party in	6
of each word	6
to generate the	6
of twitter and	6
through the lens	6
association for the	6
in all the	6
of the spread	6
here we see	6
we obtain the	6
to engage in	6
of news and	6
in the city	6
figure shows that	6
sources such as	6
to collect data	6
a political news	6
the benefits of	6
important to note	6
for each day	6
as a news	6
social network sites	6
the paper is	6
the sentiment classification	6
we counted the	6
and the social	6
of confirmed cases	6
of the nodes	6
is essential to	6
that did not	6
and understanding of	6
on the one	6
one or more	6
of a public	6
sentiment of the	6
for the same	6
understand the chain	6
other altmetric data	6
data and the	6
the potential for	6
in the face	6
the similarity between	6
on the political	6
semantic role labeling	6
and fake news	6
across all the	6
the most commonly	6
to may th	6
similar research topics	6
the users who	6
in this way	6
more than times	6
detection on twitter	6
of facebook use	6
in the table	6
public health surveillance	6
a method to	6
in the absence	6
different political parties	6
to model the	6
by the fact	6
peaks of moc	6
to see how	6
related to this	6
hashtag groups and	6
by the users	6
number of nodes	6
of the dataset	6
based on this	6
right wing community	6
comparison of the	6
an early warning	6
untrue political stories	6
during hurricane sandy	6
of tweets is	6
to address this	6
a higher likelihood	6
would be more	6
the very first	6
reported a greater	6
a need to	6
information regarding the	6
and intensity of	6
could be a	6
fields and research	6
and of the	6
in the middle	6
given that the	6
the extent of	6
a survey on	6
figure and figure	6
team brand equity	6
bigru and cnn	6
the core of	6
due to a	6
there may be	6
of a resilient	6
the most prominent	6
was able to	6
owing to the	6
six main categories	6
hundreds of thousands	6
of the coronavirus	6
confirmed cases in	6
in some cases	6
of the brand	6
weaker relationship between	6
the most used	6
which they are	6
in each of	6
our goal is	6
and quote tweets	6
this can be	6
different social media	6
the virus is	6
the uncertainty of	6
data for this	6
used in social	6
of data presence	6
are positively related	6
period of data	6
did not have	6
the attention of	6
like to thank	6
the semantic networks	6
toward uk mps	6
presented in the	6
proportion of the	6
between these two	6
was carried out	6
assets are positively	6
and the average	6
data sources are	6
we developed a	6
between democrats and	6
facebook and twitter	6
for further analysis	6
attributed to source	6
the term frequency	6
communities in large	6
the names of	6
associated with self	6
well as a	6
the tweet contains	6
the prevalence of	6
new york city	6
conflicts of interest	6
network of the	6
presented in this	6
over the world	6
abuse toward mps	6
and sentiment analysis	6
of stakeholder value	6
we analyzed the	6
they are not	6
in major league	6
from search were	6
measles outbreak in	6
a portion of	6
the city of	6
has not been	6
is lower than	6
positive for covid	6
of novel coronavirus	6
of data and	6
rather than facebook	6
the dataset is	6
analysis of data	6
using twitter and	6
american society of	6
to this end	6
of the sample	6
a description of	6
a weaker relationship	6
reported in the	6
and information diffusion	6
times higher than	6
statistics of the	6
belief that the	6
public health agencies	6
indicates that the	6
tended to be	6
the hypothesis that	6
an alternative to	6
in the top	6
story online that	6
that the majority	6
seeded lda model	6
order to understand	6
for users who	6
in the nyt	6
probability of observing	6
regard to the	6
of profanity words	6
social media by	6
receiving a high	6
the current work	6
the real network	6
to be seen	6
for the following	6
harmful media coverage	6
of tweets from	6
interested in the	6
was followed by	6
this paper presents	6
control and prevention	6
is useful for	6
we present our	6
the pandemic on	6
the death nls	6
policy makers and	6
to check the	6
the situation in	6
addition to that	6
to public health	6
missile test surge	6
the streaming api	6
in contact with	6
chain of events	6
the network and	6
in consideration of	6
to classify the	6
score for each	6
briefing on covid	6
of the web	6
the web page	6
for both the	6
news story online	6
for all tweets	6
remarks at the	6
the face of	6
the twitter stream	6
from the dataset	6
these tweets were	6
if the account	6
which do not	6
example of a	6
to address the	6
number of unique	6
from further analysis	6
compared with the	6
that many people	6
to filter out	6
who did not	6
based on different	6
composed by the	6
a corpus of	6
in contrast to	6
we would like	6
the intensity of	6
measure of the	6
presented in table	6
into six main	6
seriousness of the	6
public health crises	6
public health and	6
the media briefing	6
detection and tracking	6
of big data	6
cloud in fig	6
this measure is	6
the framing of	6
with madrid and	6
each pair of	6
mental health problems	6
used by the	6
to be an	6
according to their	6
cwts classification system	6
the progression of	6
in the original	6
an original tweet	6
generated by the	6
pneumonia in wuhan	6
the source of	6
of information diffusion	6
the understanding of	6
media data sources	6
in the process	6
identify hot research	6
performance of our	6
tweets based on	5
the two most	5
we applied the	5
presence of different	5
for our model	5
a timely manner	5
is dedicated to	5
authoritativeness of the	5
the lockdown period	5
was not certified	5
by the twitter	5
it is labeled	5
connected to the	5
the differences in	5
retweets and favorites	5
each tweet was	5
risk perception and	5
the data sources	5
and political frame	5
and health sciences	5
is likely that	5
the right wing	5
study found that	5
the news media	5
tweets from march	5
a content analysis	5
explained by the	5
have been conducted	5
the pew research	5
sharing of untrue	5
users and the	5
retrieved from the	5
was done on	5
importance of the	5
media platforms such	5
data as social	5
information and communication	5
alpha for this	5
a linguistic analysis	5
those who have	5
analysis and the	5
to look at	5
public health organizations	5
and use of	5
the first and	5
and higher levels	5
participant demographics are	5
is followed by	5
to test the	5
the results for	5
the communication ecosystem	5
of the infection	5
to collect and	5
by the disease	5
the different methods	5
we study the	5
of tweets containing	5
found out was	5
hall of fame	5
and do not	5
big data from	5
were also more	5
to detect depression	5
german speaking twitter	5
most common symptoms	5
we observed that	5
of data for	5
people were more	5
of the daily	5
between citations and	5
this suggests that	5
in social big	5
list of words	5
average number of	5
the title of	5
most commonly used	5
is the total	5
an important issue	5
done on twitter	5
over the course	5
asked to rate	5
model with the	5
ranged from to	5
to tweet about	5
of twitter messages	5
of tweets by	5
on twitter about	5
in figure and	5
set of users	5
for the surgery	5
based on our	5
social media twitter	5
and the peaks	5
machine learning on	5
a selection of	5
data sources for	5
in the geocov	5
social media metrics	5
to control the	5
smith et al	5
not certified by	5
within the network	5
online engagement and	5
the school closures	5
social learning and	5
this was done	5
better understanding of	5
by peer review	5
of information to	5
surgery of trauma	5
to be able	5
the social web	5
to slow the	5
had seen them	5
of replies received	5
from the center	5
home save lives	5
were collected from	5
this is that	5
reporting a higher	5
of which are	5
the analysis results	5
fast unfolding of	5
of using social	5
political polarization on	5
of communities in	5
have more than	5
contents of the	5
patterns of information	5
the original tweet	5
tweets originating from	5
news and the	5
set of publications	5
in this research	5
can find the	5
in the em	5
methods used by	5
of the relevant	5
can be accessed	5
is the set	5
the tweets collected	5
is clear that	5
this could be	5
algorithm to detect	5
influenced by the	5
no conflict of	5
rise of social	5
increase the impact	5
and support vector	5
in which each	5
the results presented	5
coverage on the	5
later found out	5
as far as	5
in each panel	5
stated that they	5
quantitatively and qualitatively	5
in the philippines	5
that they have	5
on the whole	5
consistent with their	5
not only the	5
between the death	5
involved in the	5
detection from twitter	5
to contribute to	5
title of the	5
to be shared	5
the control group	5
the detailed information	5
the contents of	5
the twitter posts	5
and big data	5
who do not	5
there are many	5
there are micro	5
notation description see	5
we considered the	5
of our study	5
media in the	5
to establish whether	5
analysis and opinion	5
we have developed	5
but do not	5
social network data	5
distribution with a	5
a very skewed	5
considerable amount of	5
whether the same	5
cough and fever	5
of marketing assets	5
well as for	5
we remove the	5
results from the	5
a platform for	5
of the opposition	5
true and false	5
have developed a	5
the real world	5
most frequent words	5
in our data	5
is focused on	5
in a timely	5
public opinion and	5
from the perspective	5
had a very	5
stay home save	5
the difference between	5
on advances in	5
analysis and topic	5
a very large	5
be an important	5
set of features	5
of the algorithm	5
a personal negative	5
a link between	5
the personal versus	5
identifying hot research	5
sections of the	5
known as the	5
machine learning model	5
a period of	5
for data collection	5
not only to	5
representation of a	5
related stress symptoms	5
to the original	5
the second most	5
in table are	5
in this area	5
january to april	5
and analysis of	5
as a proxy	5
natural language toolkit	5
media data mining	5
to assign a	5
at this stage	5
networks can be	5
between tweets and	5
of tweets for	5
social networks have	5
twitter data and	5
quantitative analysis of	5
of the studies	5
korea missile test	5
likely to engage	5
that it does	5
this time period	5
a sequence of	5
law enforcement agencies	5
big data analysis	5
the population of	5
outbreak in the	5
neural networks and	5
but does not	5
of twitter accounts	5
working hard to	5
number of twitter	5
of tweets to	5
events related to	5
that twitter can	5
patterns in the	5
these types of	5
the bipartite directed	5
the trend of	5
the speed of	5
of the cases	5
the propagation of	5
is composed of	5
of the media	5
data can be	5
data for research	5
wide number of	5
the th of	5
issues such as	5
male or female	5
share of the	5
results of this	5
power to detect	5
the different hashtag	5
to access the	5
used to study	5
will need to	5
twitter sentiment classification	5
network analysis to	5
row of panels	5
have a higher	5
about the covid	5
to keep the	5
a collection of	5
during the pre	5
large amounts of	5
english language tweets	5
high number of	5
included in this	5
a predictor variable	5
on the platform	5
to illustrate the	5
they are likely	5
be useful for	5
subset of the	5
the accuracy assessment	5
on the right	5
of h b	5
was supported by	5
were found to	5
word cloud on	5
any data analysis	5
a network of	5
data in a	5
the training dataset	5
to describe the	5
appearing in the	5
the information ecology	5
a wide number	5
the big data	5
to our knowledge	5
of artificial intelligence	5
have been designed	5
effect of social	5
pairs of words	5
the numbers of	5
but it is	5
of social learning	5
also more likely	5
can be observed	5
history of the	5
respect to the	5
summary of the	5
twitter data can	5
the cwts classification	5
the tweet ids	5
with the following	5
and the use	5
of the weber	5
in figure a	5
considered as a	5
the coronavirus topic	5
for this analysis	5
are included in	5
is vital to	5
collected twitter data	5
can be difficult	5
linked to the	5
the pattern of	5
was assigned to	5
media coverage and	5
states during the	5
a tweet was	5
of an outbreak	5
paper is organized	5
consistent with our	5
a specific disease	5
frames used by	5
in understanding the	5
information provided by	5
the extension of	5
excluded from further	5
the previous subsection	5
between political parties	5
in both the	5
deep learning for	5
conference on advances	5
the research topic	5
we can also	5
been widely used	5
increasing use of	5
can be described	5
during the lockdown	5
the tweets into	5
has been shown	5
titles of the	5
that they were	5
most of these	5
we collected all	5
to characterize the	5
of altmetric events	5
magnitude of the	5
to show that	5
false political stories	5
from the user	5
they can be	5
point in time	5
has shown that	5
which is an	5
to examine how	5
networks have become	5
of the best	5
small number of	5
on the map	5
it remains to	5
tweets per day	5
illustrated in figure	5
exposure to the	5
for early detection	5
also used to	5
to see the	5
for people to	5
levels of social	5
the news coverage	5
on twitter using	5
the resulting network	5
does not contain	5
the time was	5
table and fig	5
cloud on the	5
and showed that	5
could also be	5
of the situation	5
females were more	5
section of the	5
a matter of	5
unfolding of communities	5
features of twitter	5
sum of the	5
of different altmetric	5
has become a	5