=Paper=
{{Paper
|id=Vol-2890/paper4
|storemode=property
|title=An Analysis of People's Reasoning for Sharing Real and Fake News
|pdfUrl=https://ceur-ws.org/Vol-2890/paper4.pdf
|volume=Vol-2890
|authors=Anu Shrestha,Francesca Spezzano
|dblpUrl=https://dblp.org/rec/conf/www/ShresthaS21
}}
==An Analysis of People's Reasoning for Sharing Real and Fake News==
https://ceur-ws.org/Vol-2890/paper4.pdf
An Analysis of People’s Reasoning for Sharing
Real and Fake News
Anu Shrestha and Francesca Spezzano
Computer Science Department
Boise State University
anushrestha@u.boisestate.edu
francescaspezzano@boisestate.edu
Abstract. The problem of the increase in the volume of fake news and
its widespread over social media has gained massive attention as most of
the population seeks social media for daily news diet. Humans are equally
responsible for the surge of fake news spread. Thus, it is imperative to
understand people’s behavior when they decide to share real and fake
news items on social media. In an attempt to do so, we performed an
analysis on data collected through a survey where participants (n= 363
) were asked whether they were willing to share the given news item on
their social media and explain the reasoning for their decision. The re-
sults show that the analysis presents several commonalities with previous
studies. Moreover, we also addressed the problem of predicting whether
a person will share a given news item or not. For this, we used intrin-
sic features from participants’ open-ended responses and demographics
attributes. We found that the perceived emotions triggered by the news
item show a strong influence on the user’s decision to share news items
on social media.
Keywords: Fake News · News Sharing · Emotion · Misinformation ·
Social Media.
1 Introduction
Social media has emerged as popular information source people rely on for events,
breaking news, and emergencies. Indeed, it has become a source of daily news
diet for the increasingly large population. Statistics show that majority of the
population (71% of American adults) ever get news through social media in
2020 [24] which was increased by 3% since 2018 [23]. The landscape of news
consumption and information flow has drastically changed with the popularity
of social media. It has transformed how news content is created, how people en-
gage with news items, and share information, blurring the journalists’ boundary
Copyright © 2021 for this paper by its authors. Use permitted under Creative Com-
mons License Attribution 4.0 International (CC BY 4.0). Presented at the MISINFO
2021 Workshop, held in conjunction with the 30th ACM The Web Conference, 2021,
in Ljubljana, Slovenia.
2 A. Shrestha and F. Spezzano
in traditional media that is first verifying and then disseminating only the ac-
curate news items [25]. Moreover, users in social media (both organizations and
individuals) actively participate in creating and sharing news items with friends,
families, and other readers due to its ease of use, lower cost, and convenience
of further sharing [2, 29]. This shift of news paradigm has led to an unprece-
dented transformation in both news quality and quantity that users encounter
in social media, increasing the probability of potential encounters and the spread
of fake news, fostering social media as a fertile ground for the production and
propagation of fake news.
The sheer volume of fake news being observed in social media has recently be-
come an obvious cause of concern. Many studies have highlighted the character-
istics of fake news through linguistic and psychological attributes [11, 20, 21, 27],
writing styles [3, 11, 13], network-based attributes [7] and hybrid attributes con-
sidering both linguistic and network [29].
Despite several studies illustrating cues to identify fake news and mitigate
its spread, there is a worrisome amount of fake news widely spreading over so-
cial media. Fake news has been identified as more likely to go viral than real
news, spreading faster and wider [35]. Additionally, an analysis of news about
the 2016 election conducted by BuzzFeed, also found more engagement with fake
news than real news [32]. Earlier studies analyzed the potential reason behind
this rapid diffusion of news in social media, focusing on various factors, in-
cluding polarized communities of users with common belief (echo-chambers) [4],
epidemiological models [12]. Some studies highlighted the actors responsible for
spreading fake news, including bots and cyborgs [6]. Although bots are equally
responsible for spreading real and fake news, the considerable spread of fake
news is caused by human activity [30, 35] as people are generally not able to
accurately identify which news item is fake and which is real [34]. Thus, it is
crucial to understand the people’s sharing behavior of fake and real news on
social media to minimize fake news diffusion.
In this context, this study seeks to better understand how people reason
when they decide to share real news and fake news. In particular, we surveyed
363 undergraduate students where we asked participants to report and explain
their willingness to share given news item (with headline and image) on their
social media. We also leveraged the demographic attributes of participants like
gender and political orientation in our study. We performed a comprehensive
data analysis to investigate the pattern of news sharing behavior, the role of
demographics in news sharing decisions, and why people share real and fake
news. Furthermore, we addressed the problem of predicting whether a person
will share a given news item or not according to emotion, psychological, and
demographics features as a binary classification task.
Our experiments show several commonalities with previous findings regarding
news sharing behavior.
– News sharing is rare as only a small percentage (19.2% to 28.2%) of users
expressed the willingness to share news in social media, regardless of news
veracity.
An Analysis of People’s Reasoning for Sharing Real and Fake News 3
– Female participants are prone to share more news than male participants
regardless of news veracity.
– Left-leaning participants tend to share real news more than fake news, inde-
pendently of the news source’s political orientation, and right-leaning par-
ticipants were instead more prone to share news items from sources with the
same political-leaning, independently of news veracity.
– The prominent themes illustrated by the approaches used by participants to
make their sharing decisions falls under subjectivity and the focus on others
interest or disinterest in news topic.
– Emotion features are more effective in predicting people’s willingness to share
a given news item.
2 Related Work
Several studies have been conducted to understand the characteristics of users
that are likely to contribute to spreading fake news on social networks. Vosoughi
et al. [35] revealed that the fake news spreaders had, on average, significantly
fewer followers, followed significantly fewer people, and were significantly less
active on Twitter. Moreover, bots tend to spread both real and fake news, and
the considerable spread of fake news on Twitter is caused by human activity.
Shrestha and Spezzano showed that social network properties help in identifying
active fake news spreaders [26]. Shu et al. [30] analyzed user profiles to under-
stand the characteristics of users that are likely to trust/distrust fake news. They
found that, on average, users who share fake news tend to be registered for a
shorter time than the ones who share real news and that bots are more likely
to post a piece of fake news than a real one, even though users who spread
fake news are still more likely to be humans than bots. They also show that
real news spreaders are more likely to be more popular and that older people
and females are more likely to spread fake news. Guess et al. [9] also analyzed
user demographics as predictors of fake news sharing on Facebook and found
out political-orientation, age, and social media usage to be the most relevant.
Specifically, people are more likely to share articles they agree with (e.g., right-
leaning people tended to share more fake news because the majority of the fake
news considered in the study were from 2016 and pro-Trump), seniors tend to
share more fake news probably because they lack digital media literacy skills
that are necessary to assess online news truthfulness, and the more people post
in social media, the less they are likely to share fake news, most likely because
they are familiar with the platform and they know what they share.
Shrestha et al. [28] analyzed the linguistic patterns used by a user in their
tweets and personality traits as a predictor for identifying users who tend to
share fake news on Twitter data [22, 28]. Likewise, Giachanou et al. [8] pro-
posed an approach based on a convolutional neural network to process the user
Twitter feed in combination with features representing user personality traits
and linguistic patterns used in their tweets to address the problem of discrimi-
nating between fake news spreaders and fact-checkers.
4 A. Shrestha and F. Spezzano
Fig. 1: News items used in our survey instrument.
Ma et al. [15] went beyond the user and news characteristics and analyzed
the characteristics of diffusion networks to explain users’ news sharing behavior.
They found opinion leadership, news preference, and tie strength to be the most
important factors at predicting news sharing, while homophily hampered news
sharing in users’ local networks. Also, people driven by gratifications of infor-
mation seeking, socializing, and status-seeking were more likely to share news
on social media platforms [14].
3 Data Collection
We conducted an online survey delivered via Qualtrics. Through this online sur-
vey, participants were given four news headlines and accompanying images. For
each news item, participants were asked whether they were willing to share the
given news item on their social media and write an explanation of the reasoning
for their decision. We considered the four news items shown in Figure 1 and
gathered from politifact.com. In this news set, two are real news items, and
two are fake news items, as fact-checked by politifact.com. Both real and fake
news items are one from a left-leaning source and one from a right-leaning source.
News source political-leaning has been gathered from mediabiasfactcheck.com.
We recruited undergraduate students (n = 363 ) from a volunteer pool in
general education social science courses (Psychology 101) to participate in our
survey (258 F, 101 M, 4 Other; mean age 19.7, SD = 4.25). The research was ap-
proved by the university IRB. Participants were compensated with course credit
(volunteering for studies being one option for a research experience requirement).
Participants received no training.
An Analysis of People’s Reasoning for Sharing Real and Fake News 5
Percentage of Sharing
News Item 1 (Fake) 19.2%
News Item 2 (Fake) 22.9%
News Item 3 (Real) 20.0%
News Item 4 (Real) 28.2%
Table 1: News Sharing Behavior.
4 Data Analysis
News sharing is rare. We start the analysis of our data by observing that only
a small percentage of users expressed the willingness to share news in social
media, independently of the veracity of the news. As shown in Table 1, this
percentage ranges between 19.2% and 28.2% among the news considered in our
survey. Previous research [10] has shown that sharing news articles from fake
news domains on Facebook was a rare activity during the 2016 U.S. presidential
campaign. Our data on fake news sharing is aligned with this result, but our
respondents also showed some preliminary evidence that this pattern may be
true for real news sharing as well.
The role of demographics in news sharing. We collected demographic data from
our survey participants, including gender, political orientation, and age. As most
participants are in the same age range (18-25), we did not consider age in our
analysis.
When looking at differences in sharing behavior according to gender (see
Figure 2), we observe that the female participants were more prone to share
both the fake news items considered than male participants who were more
skeptical about the same news items. Shu et al. [31] in his studies have shown a
similar result where female users tend to trust fake news more than male users.
In general, females were more prone to share more news items than males (three
vs. one).
Regarding participants political orientation, we see two interesting patterns
as reported in Figure 3: (1) left-leaning participants were more prone to share
real news than fake news, independently of the political orientation of the news
source; (2) right-leaning participants were instead more prone to share news
items from sources with the same political-leaning (news items 1 and 3), in-
dependently of news veracity. Similarly, Guess et al. [10] have shown that, in
2016, conservatives were more likely to share articles from pro-Trump fake news
domains than liberals or moderates.
Why people share real and fake news? Yaqub et al. [36] analyzed open-ended
responses of participants in the study where they explained the reason behind
their intention to share true, false, and satire headlines. In their study, the most
frequent rationales behind sharing/not sharing news were (1) the interest/non-
interest towards the news, (2) the potential of generating discussion among the
6 A. Shrestha and F. Spezzano
(a) News item 1 (fake) (b) News item 2 (fake)
(c) News item 3 (real) (d) News item 4 (real)
Fig. 2: Distribution of participant’s gender.
friends, (3) the fact that the news is not relevant to the user’s life, and (4) the
perceived news credibility, especially as a motivation for not sharing news.
We conducted a similar analysis on a sample of our data (n=25). Specifically,
we conducted a thematic analysis to identify the prominent themes that illus-
trated the approaches used by participants to make their sharing decisions. We
followed an inductive approach to generating codes [5]. We found out the prin-
cipal codes to be focused on potential others (”My friends would/would not be
interested in this”), interest or disinterest in the news topic, and subjectivity/the
self (”I would/wouldn’t share this because...”, ”I would call that fake/real”) and
are mostly aligned with the finding by Yaqub et al. [36].
Regarding performing credibility assessment before making the sharing de-
cision, we also found in our sample data that this was performed more often for
fake news (28% of the times for news item 1 and 56% for news item 2) than for
real news (24% of the times for news item 3 and 16% for news item 4). Moreover,
when performed, the credibility assessment was much more correct in the case
of fake news (100% of the times for news item 1 and 93% for news item 2) than
real news (67% of the times for news item 3 and 25% for news item 4).
Overall, the data analysis performed in this section shows that our collected
data presents several commonalities with previous studies, ensuring we have qual-
ity data suitable for further investigations.
An Analysis of People’s Reasoning for Sharing Real and Fake News 7
(a) News item 1 (b) News item 2
(fake, right-leaning source) (fake, left-leaning source)
(c) News item 3 (d) News item 4
(real, right-leaning source) (real, left-leaning source)
Fig. 3: Distribution of participant’s self-identified political orientation.
5 Predicting News Sharing
In this section, we address the problem of predicting whether a person will share
or not a news item according to emotion and psychological features generated
when they consider a news item and demographics (gender and political orien-
tation) as well. We modeled the problem as a binary classification task where
we computed emotion and psychological features from participants’ open-ended
responses to the survey question asking for an explanation of their decision to
share or not the given news item.
5.1 Textual Features Extraction
Emotion Features (Emotion) In order to compute a vector of scores quan-
tifying participants’ emotions when deciding whether or not to share a news
item, we considered their open-ended survey responses and proceeded as fol-
lows. We started by cleaning responses’ text by expanding contraction words,
correcting misspellings and grammatical mistakes using LanguageTool1 and re-
placing negated words with their WordNet antonym. Next, we extracted emo-
1
https://pypi.org/project/language-tool-python/
8 A. Shrestha and F. Spezzano
tions from the text by using the Emotion Intensity Lexicon (NRC-EIL) [18]
and EmoLex [33]. Emotion features computed via NRC-EIL include anger, joy,
sadness, fear, disgust, anticipation, surprise, and trust, while Emolex2 features
include happy, sad, angry, don’t care, inspired, afraid, amused, and annoyed. Fea-
ture vectors have been computed by using the approaches proposed in [16,17]. In
addition, we also considered emotion-related features as computed by the 2015
Linguistic Inquiry and Word Count (LIWC) [19] tool, which includes effective
processes like anxiety, anger, positive and negative emotion.
Psycho-linguistic Features (LIWC) To understand the relationship between
psychological states and the participants’ decision-making, we considered the set
of psycho-linguistic features computed by the Linguistic Inquiry and Word Count
(LIWC) tool [19]. LIWC is a transparent text analysis tool that counts words in
psychologically meaningful categories. Specifically, we considered psychological
processes that include social processes (e.g., family, friends), cognitive processes
(e.g., think, cause, perhaps), perceptual processes (e.g., see, heard, felt), biologi-
cal processes (e.g., eat, pain, love), relativity (e.g., area, move, day) and personal
concerns (e.g., work, leisure, achieve, home, money, religion, death).
Demographics (Demog) As explicit features, we used participants’ self-identified
gender and political orientation to understand if the demographic attributes pro-
vide potential cues in predicting users’ sharing decisions.
5.2 Experimental Setting and Results
We used each group of features described in the previous section as input to
a random forest classifier to compute the performance of these features in pre-
dicting whether a reader of a news item (a participant of our survey) is willing
to share or not the given news item on their social networks. We also tried
other classifiers such as Support Vector Machine (SVM) and logistic regression,
but random forest achieved the best results. Hence, in the paper, we report the
results of random forest only. We used class weighting to deal with the class
imbalance and performed 5-fold cross-validation.
The results are reported in Table 2 according to the area under the ROC
curve (AUROC), average precision (AvgP), and F1-measure (F1). As can be
seen, when each feature group is considered separately, emotion features are the
best performing features compared to LIWC features and demographics with
72% vs. 61% and 52% AUROC and 40% vs. 25% and 20% average precision
for news item 1, 71% vs. 61% and 57% AUROC and 42% vs. 31% and 25%
average precision for news item 2, 77% vs. 59% and 62% AUROC and 58%
vs. 31% and 26% average precision for news item 3 and 78% vs. 61% and 59%
AUROC and 56% vs. 40% and 42% average precision for news item 4 (bold
in Table 2). We further considered a combination of all feature groups to see
2
https://sites.google.com/site/emolexdata/
An Analysis of People’s Reasoning for Sharing Real and Fake News 9
Features AUROC AvgP F1
LIWC 0.611 0.247 0.166
News Item 1 (Fake) Demog 0.518 0.207 0.228
Emotion 0.720 0.403 0.228
All 0.722 0.382 0.129
LIWC 0.608 0.307 0.175
News Item 2 (Fake) Demog 0.565 0.250 0.325
Emotion 0.706 0.416 0.162
All 0.707 0.421 0.122
LIWC 0.586 0.310 0.257
News Item 3 (Real) Demog 0.617 0.258 0.300
Emotion 0.771 0.578 0.477
All 0.796 0.585 0.439
LIWC 0.611 0.397 0.302
News Item 4 (Real) Demog 0.590 0.317 0.356
Emotion 0.784 0.564 0.423
All 0.786 0.562 0.359
Table 2: Comparison of emotion, psycho-linguistic, and demographic features to
predict whether a news item will be shared or not. We used a random forest
classifier. Best results among feature groups considered separately are in bold.
Best overall results are shaded.
if combining demographics, psychological and emotional features can provide
complementary information that can help improve the prediction. We observed
that when the combination of all feature groups is considered, the performance
remained more or less the same if not improved according to AUROC (shaded
in Table 2). This demonstrates that emotion features are more effective than
other groups of features considered in our study for predicting people’s sharing
behavior. Hence, one of the motivations for potential news-sharing behavior in
social media could be emotional persuasion. It will not be inaccurate to say that
being persuaded by strong emotions like anger, fear, surprise, joy, etc., triggered
by news content, people tend to get involved and share more news on social
media. This finding aligns with the previous research by Berger et al. [1] which
also states that emotional arousal tends to increases the likelihood of sharing
news on social media.
6 Conclusion and Future Work
To sum up, this paper presents findings from studying people’s reasoning when
they decide to share real and fake news items provided with headlines and im-
ages. This paper investigates the correlation between the user’s sharing decision
and explicit attributes provided by participants like demographics and politi-
cal orientation. Furthermore, we addressed the problem of predicting whether
a person will share a given news item or not using intrinsic features like psy-
10 A. Shrestha and F. Spezzano
chological and emotion from participants’ open-ended responses explaining their
willingness to share given news item along with demographics attributes.
The results show that news sharing is rare, and among the participants ex-
pressing willingness to share, females are prone to share more news in general.
Participants’ political orientation exerts a significant pattern on news sharing
behavior that is left-leaning participants’ news sharing behavior is motivated by
news veracity rather than political orientation. In contrast, it is the other way
around for right-leaning participants. Likewise, it shows the possibility of users
sharing news items depends on the perceived relevance of news interest among
friends and families. Moreover, this paper also highlights that the perceived emo-
tions triggered by the news item show a strong influence on user’s news sharing
behavior in social media.
One potential limitation of our study is that we have considered only four
news of each political leaning (2 fake and 2 real). Considering a bigger set of
news items could have shown significant patterns and support to our findings.
Furthermore, this work focuses on a younger sample of the limited range of age,
due to which we did not consider age in demographic attributes. It could have
added some more insights regarding news sharing behavior among different age
groups if we could consider participants of a wide range of ages (from younger
to older population). We will address these limitations in our future work.
Acknowledgements
This work has been supported by the National Science Foundation under Award
no. 1943370. We thank Brian Stone for facilitating the data collection and Ashlee
Milton and Maria Soledad Pera for providing us with the code used in their
papers [16, 17] to compute emotional features.
References
1. Berger, J.: Arousal increases social transmission of information. Psychological
science 22(7), 891–893 (2011)
2. Bruns, A., Highfield, T., Lind, R.A.: Blogs, twitter, and breaking news: The
produsage of citizen journalism. Produsing theory in a digital world: The
intersection of audiences and production in contemporary theory 80(2012), 15–32
(2012)
3. Castillo, C., Mendoza, M., Poblete, B.: Information credibility on twitter. In:
Proceedings of the 20th international conference on World wide web. pp. 675–684
(2011)
4. Del Vicario, M., Bessi, A., Zollo, F., Petroni, F., Scala, A., Caldarelli, G.,
Stanley, H.E., Quattrociocchi, W.: The spreading of misinformation online.
Proceedings of the National Academy of Sciences 113(3), 554–559 (2016)
5. DNSc, J., Strauss, A.: Basics of Qualitative Research: Techniques and Procedures
for Developing Grounded Theory. SAGE Publications, Inc, Los Angeles, fourth
edition edn. (Dec 2014)
An Analysis of People’s Reasoning for Sharing Real and Fake News 11
6. Ferrara, E., Varol, O., Davis, C., Menczer, F., Flammini, A.: The rise of social
bots. Communications of the ACM 59(7), 96–104 (2016)
7. Gayo-Avello, D., Metaxas, P.T., Mustafaraj, E., Strohmaier, M., Schoen, H.,
Gloor, P., Castillo, C., Mendoza, M., Poblete, B.: Predicting information
credibility in time-sensitive social media. Internet Research (2013)
8. Giachanou, A., Rissola, E.A., Ghanem, B., Crestani, F., Rosso, P.: The role of
personality and linguistic patterns in discriminating between fake news spreaders
and fact checkers. In: Natural Language Processing and Information Systems:
25th International Conference on Applications of Natural Language to
Information Systems, NLDB 2020. p. 181. Springer Nature
9. Guess, A., Nagler, J., Tucker, J.: Less than you think: Prevalence and predictors
of fake news dissemination on facebook. Science advances 5(1), eaau4586 (2019)
10. Guess, A., Nagler, J., Tucker, J.: Less than you think: Prevalence and predictors
of fake news dissemination on facebook. Science Advances 5(1) (2019)
11. Horne, B.D., Adali, S.: This just in: Fake news packs a lot in title, uses simpler,
repetitive content in text body, more similar to satire than real news. In: The 2nd
International Workshop on News and Public Opinion at ICWSM (2017)
12. Jin, F., Dougherty, E., Saraf, P., Cao, Y., Ramakrishnan, N.: Epidemiological
modeling of news and rumors on twitter. In: Proceedings of the 7th workshop on
social network mining and analysis. pp. 1–9 (2013)
13. Jin, Z., Cao, J., Zhang, Y., Luo, J.: News verification by exploiting conflicting
social viewpoints in microblogs. In: Proceedings of the AAAI Conference on
Artificial Intelligence. vol. 30 (2016)
14. Lee, C.S., Ma, L.: News sharing in social media: The effect of gratifications and
prior experience. Computers in human behavior 28(2), 331–339 (2012)
15. Ma, L., Lee, C.S., Goh, D.H.: Understanding news sharing in social media from
the diffusion of innovations perspective. In: 2013 IEEE International Conference
on Green Computing and Communications and IEEE Internet of Things and
IEEE Cyber, Physical and Social Computing. pp. 1013–1020. IEEE (2013)
16. Milton, A., Batista, L., Allen, G., Gao, S., Ng, Y., Pera, M.S.: ”don’t judge a
book by its cover”: Exploring book traits children favor. In: RecSys 2020:
Fourteenth ACM Conference on Recommender Systems, Virtual Event, Brazil,
September 22-26, 2020. pp. 669–674. ACM (2020)
17. Milton, A., Pera, M.S.: What snippets feel: Depression, search, and snippets. In:
Cantador, I., Chevalier, M., Melucci, M., Mothe, J. (eds.) Proceedings of the
Joint Conference of the Information Retrieval Communities in Europe (CIRCLE
2020), Samatan, Gers, France, July 6-9, 2020. CEUR Workshop Proceedings,
vol. 2621 (2020)
18. Mohammad, S.: Word affect intensities. In: Proceedings of the Eleventh
International Conference on Language Resources and Evaluation, LREC 2018,
Miyazaki, Japan, May 7-12, 2018. European Language Resources Association
(ELRA) (2018)
19. Pennebaker, J.W., Boyd, R.L., Jordan, K., Blackburn, K.: The development and
psychometric properties of liwc2015. Tech. rep. (2015)
20. Pérez-Rosas, V., Kleinberg, B., Lefevre, A., Mihalcea, R.: Automatic detection of
fake news. In: Proceedings of the 27th International Conference on
Computational Linguistics. pp. 3391–3401 (2018)
21. Potthast, M., Kiesel, J., Reinartz, K., Bevendorff, J., Stein, B.: A stylometric
inquiry into hyperpartisan and fake news. In: Proceedings of the 56th Annual
Meeting of the Association for Computational Linguistics (Volume 1: Long
Papers). pp. 231–240 (2018)
12 A. Shrestha and F. Spezzano
22. Rangel, F., Giachanou, A., Ghanem, B., Rosso, P.: Overview of the 8th author
profiling task at pan 2020: Profiling fake news spreaders on twitter. In: CLEF
(2020)
23. Shearer, E., Matsa, K.E.: News use across social media platforms 2018. Pew
Research Center (2018)
24. Shearer, E., Mitchell, A.: News use across social media platforms in 2020. Pew
Research Center (2020)
25. Shoemaker, P.J., Vos, T.P., Reese, S.D.: Journalists as gatekeepers. The
handbook of journalism studies 73 (2009)
26. Shrestha, A., Spezzano, F.: Online misinformation: from the deceiver to the
victim. In: ASONAM ’19: International Conference on Advances in Social
Networks Analysis and Mining. pp. 847–850. ACM (2019)
27. Shrestha, A., Spezzano, F.: Textual characteristics of news title and body to
detect fake news: A reproducibility study. In: Advances in Information Retrieval -
43rd European Conference on IR Research, ECIR 2021, Proceedings, Part II.
Lecture Notes in Computer Science, vol. 12657, pp. 120–133. Springer (2021)
28. Shrestha, A., Spezzano, F., Joy, A.: Detecting fake news spreaders in social
networks via linguistic and personality features. In: CLEF (2020)
29. Shu, K., Sliva, A., Wang, S., Tang, J., Liu, H.: Fake news detection on social
media: A data mining perspective. ACM SIGKDD explorations newsletter 19(1),
22–36 (2017)
30. Shu, K., Wang, S., Liu, H.: Understanding user profiles on social media for fake
news detection. In: 1st IEEE International Workshop on Fake MultiMedia
(FakeMM 2018) (2018)
31. Shu, K., Wang, S., Liu, H.: Understanding user profiles on social media for fake
news detection. In: 2018 IEEE Conference on Multimedia Information Processing
and Retrieval (MIPR). pp. 430–435. IEEE (2018)
32. Silverman, C.: This analysis shows how viral fake election news stories
outperformed real news on facebook. BuzzFeed News 16 (2016)
33. Song, K., Gao, W., Chen, L., Feng, S., Wang, D., Zhang, C.: Build emotion
lexicon from the mood of crowd via topic-assisted joint non-negative matrix
factorization. In: Proceedings of the 39th International ACM SIGIR conference
on Research and Development in Information Retrieval, SIGIR 2016. pp.
773–776. ACM (2016)
34. Spezzano, F., Shrestha, A., Fails, J.A., Stone, B.W.: That’s fake news!
investigating how readers identify the reliability of news when provided title,
image, source bias, and full articles. Proceedings of the ACM on Human
Computer Interaction journal 5(CSCW1, Article 109) (2021)
35. Vosoughi, S., Roy, D., Aral, S.: The spread of true and false news online. Science
359(6380), 1146–1151 (2018)
36. Yaqub, W., Kakhidze, O., Brockman, M.L., Memon, N., Patil, S.: Effects of
credibility indicators on social media news sharing intent. In: Proceedings of the
2020 CHI Conference on Human Factors in Computing Systems. p. 1–14. CHI ’20
(2020)