Defining Fake News For Algorithmic Deception Detection

UPPSALA
UNIVERSITET
Defining fake news for algorithmic deception detection

purposes
Yohannes Kassahun Asresu
Subject: Information Systems

Corresponds to: 30 hp
Presented: VT 2018
Supervisor: Claes Thorén
Examiner: Steve McKeever
Department of Informatics and Media

Abstract
One of the challenges of the digital age is the dissemination of false information on social media. False
information on social media is causing problems. Technology-based solutions are being utilized to combat
the fake news epidemic and deception detection algorithms are helping to identify false information. The
aim of the thesis is to examine how those solutions define false information. Several of them use the
ambiguous and overly misused ‘fake news’ to explain the situation. Our investigation shows that algorithmic
deception detection papers define ‘fake news’, inadequately. Most of the algorithms are developed since
2017 and the rush to put the fire out might be the reason for the unsatisfactory problem definitions.
Algorithms have a potential in combating fake news, but the lack of proper problem definition can lead to
wrong, insufficient and erroneous results.
The 16 papers we investigate show that ambiguous and mixed meanings are used to define fake news. By
utilizing thematic analysis, the thesis argues that fake news is an inadequate and improper term for academic
purposes. Our analysis demonstrates that the false information dissemination phenomena need to be
addressed and defined as disinformation.
Keywords: Fake news, Disinformation, Deception detection, Algorithm, Misinformation, Information

disorder, Meaning,
1
Table of Contents
1 Introduction 6
1.1 Background 6
1.2 Research problem 7
1.3 Research question 8
1.4 Aim of the study 8
1.5 Limitation 8
2 What is Fake News? 9

2.1 Fake news- to use it or not 9
2.2 Fake news, post-truth and gaudy facts 10
2.3 Scholarly definitions of fake news, disinformation, misinformation, propaganda
and hoax 11
2.3.1 Defining fake news 11
2.3.2 Typology of fake news 12
2.3.3 Disinformation 14
2.3.4 Misinformation 15
2.3.5 Propaganda 16
2.3.6 Hoax 16
3 Theory 17
3.1 Meaning 17
3.1.1 Semantics 17
3.1.2 Pragmatics 18
3.2 Deception detection 20
3.3 Related work 20
4 Methodology 22
4.1 Research strategy 22
4.1.1 Mapped definitions 22
4.2 Coding and analysis of definitions 26
2
4.2.1 Methods of thematic analysis 27
4.2.2 Doing the analysis 27
5 Results 33
5.1 Identified themes 33
5.2 Disinformation 33
5.2.1 Deliberate fabrication 34
5.2.2 Intentionally false 34
5.2.3 Digital deception 35
5.2.4 Falsification 35
5.2.5 Misleading 35
5.3 Misinformation 35
5.3.1 Honest mistake 36
5.3.2 Wrong information 36
5.3.3 Incorrect facts 36
5.3.4 Misleading 36
5.4 Definitions from the papers 36
6 Discussion 41
7 Conclusion 42
7.1 Future research 42
8 Bibliography 43
3
List of Figures
1 Typology of misinformation and disinformation 13
2 Final Thematic map 32
4
List of tables
1 Publication details 23
2 Six phase thematic analysis frame work 27
3 Initial codes 28
4 Searching for themes 31
5 Reviewed themes 31
6 Paper 1 Definition of sub-themes 36
5
1 Introduction
The misuse of the term ‘fake news’ has created ambiguity and confusion. Since it is election year in Sweden
we were originally planning to conduct a study about how fake news can affect the 2018 election. For that
purpose, we came across a lot of information in relation to elections and the role of fake new in swaying
votes. We read a lot of scholarly articles, journals and news reports and had an interview with the concerned
Swedish agency. We realized that the term ‘fake news’ is not adequately defined and utilized to be used as
representative term for the information disorder that we are in. There are a lot of efforts that ranges from
fact checking, news flagging, crowd sourcing and algorithm deception detection up to fake website detectors
to combat fake news and protect civil liberties as embodied in elections. The fake news epidemic might be
one of the most talked about and inadequately defined one.
The interest of the thesis shifted from studying a country’s effort in combating fake news to examining the
definitions and interpretations given to the term. Specifically, how algorithm-based deception detection
research articles define the term is the main theme of the thesis. There are hundreds of scholarly articles
with different definitions of the term. On top of that, some scholars argue against the use of ‘fake news’
since it lost all its academic meaning and purpose.
1.1 Background
fake news has become a household word since the 2016 USA election. Fake news has been named by Collins
dictionary word of the year 2017. Usage of the term has increased with 365 percent from 2016
(collinsdictionary.com). A recent CNN count found that the 45th USA President Donald Trump on average
used the term “fake news” more than once a day; 404 times between Jan 20,2017 and Jan 17, 2018 (Stelter,
2018). Even though the term gets a huge usage rate, it’s not a new phenomenon. A 1925 edition of the
Harper Magazine published an article titled “Fake news and the public”. The article indicated the emergence
of fake news (Lazar et al., 2017). Robert Darnton on his article “The true history of fake news”, has tried to
shed light on the subject from historical perspective (Darnton, 2018).
An overview of fake news
Several researchers have been studying fake news since the controversial US 2016 election. Their papers
focus on the sources of fake news, its social and political impact, and economic implications, and the type
of measures needs to be taken to combat the fake news plague.
Fake news is produced and disseminated for different purposes’ ideological, monetary and for propaganda
according to Allcott and Gentzkow, (2017). They argue that President Trump was favored by fake news
disseminators. They also provide several bench marks that show voter manipulation. The 2016 Brexit
referendum was also victim of fake news. According to news reports, intentional fake news dissemination
by the Brexiters and others played a greater role for the exit decision (Grice, 2018). Marsden (2017) has a
different view of fake news in relation to its use by some politicians; “fake news is the heartfelt cry of
politicians who feel wronged by the online media”. He continues by saying that those politicians cry for
help when they are barred from reaching their voters because of ad blocking and filter bubbles.
6
Gu et al., (2017) discuss fake news from a historical perspective. The paper focuses on how propagandists
are abusing social media to influence politics, foreign affairs and commerce. Their framework ‘the fake
news triangle’ incorporates social networks, tools and services, and the human motivation. Their study
investigates the underground market places selling tools and services for public opinion manipulation
campaigns.
The yearly report by the Reuters institute (Newman et al., 2017) focuses on fake news, changing business
models and the role of platforms. The research is based on a YouGov survey, which encompasses 36
countries. The survey participants identified lack of rules and viral algorithms as motivating reasons for the
spread of fake news.
Many businesses, public figures and celebrities are dependent on the attention they receive through social
networks. A recent New York Times investigative article reveals the complexity of fake news and the market
of tweets and likes (Confessore et al., 2018). A news report by NPR exposed one of the fake news
fabricators. With his “fake news publishing empire” he admitted that he is making hefty profits by
disseminating fake news on social media (Sydell, 2016).
The 2016 US election controversy sends a warning signal to other nations to safeguard their elections and
democratic processes from foreign interference (the local). Fake news is multifaceted, it is serving as a
propaganda machine (Gu et al., 2017), and it is being used to sway the values at stock markets (Ferrara et
al., 2016).
What is an algorithm?
Since we are going to mention algorithm several times in the thesis, it is useful to define the term. Cormen
et al., (2009) define algorithm as a computational procedure based on an input output relationship. They
also define it as a tool that can be utilized for solving a well-specified computational problem. Algorithms
take large data sets and produce correlations, interpretations, recommendations and even decisions. Smith
(2018) describe the early algorithms as logic of classical computing. The algorithms were simple, and they
are intended to perform small tasks. Smith (2018) argues that the advent of internet and search engines in
mid-1990’s changed the behavior and application of algorithms. Currently, complex algorithms are
implemented in health care, stock market, military, education system and in many more.
1.2 Research problem
Fake news has become a buzzword (Tandoc et al., 2018). Despite the common use of the term, it eludes
common definition (Wardle, 2017). The term is defined and categorized the way it suits the purpose of the
user. Lack of agreed understanding and interpretations leads the term to have conflicting definitions (Persily,
2017). Fake news is an ambiguous and misleading word (Publications.europa.eu., 2018). Because of the
misuse and abuse of the term some scholars are arguing that fake news has shifted away from academic
understanding (Haiden et al., 2018).
7
1.3 Research question
Fake news, as ambiguous as it is, it has several overlapping often mixed meanings and interpretations. It
can mean false news, misinformation, disinformation, propaganda, hoax and satire. Others put those
definitions as categories of fake news. The thesis is focuses in answering the following questions;
1. How well do algorithms define fake news?

2. What can be done to improve the definitions of fake news in algorithms that are meant to detect
fake news?
1.4 Aim of the study

Combating fake news requires a coordinated effort and algorithmic deception detection is part of the effort.
Machine learning based solutions got huge attention and soon became a discussion point. One of the major
discussions among scholars is how can algorithms understand the context, intent and the hidden words in
context. This thesis focuses on how deception detection algorithms define and understand false information.
1.5 Limitation
Fake news is multilayered. The research will only focus on how fake news is defined by algorithmic
deception detection researches. This study will not investigate how algorithms behave and function or what
methods they implement for fake news detection.
8
2 What is Fake News?
While the complexity, ambiguity and lack of common definition still surrounds fake news, there are
technological efforts to combat the dissemination of fake news. Several algorithmic solutions are developed
and proposed to solve the problem. Even if one can use the most powerful and shiniest algorithms and fail
to define the problem, as Brownlee (2018) puts it “the results will be meaningless; algorithms with the
wrong definition of fake news will ‘solve’ the wrong problem”. The aim of chapter two is to show how
definitions and meanings matter in relation to developing algorithms for fake news detection purposes.
2.1 Fake News- to use it or not
There is an on-going debate among scholars on the significance of using the term “fake news” representing
“information disorder” (Wardle, and Derakhshan, 2018). Those who oppose the use of the term cite the
term’s ambiguity and its conflicting definitions as a problem. An EU high level group on its report on
disinformation states that disinformation is the problem not fake news (Publications.europa.eu. 2018). The
group presents two main reasons for its decision to refrain from using the term “fake news”;
(1) “Firstly, the term is inadequate to capture the complex problem of disinformation, which
involves content that is not actually or completely “fake” but fabricated information blended
with facts, and practices that go well beyond anything resembling “news” to include some
forms of automated accounts used for astroturfing, networks of fake followers, fabricated or
manipulated videos, targeted advertising, organized trolling, visual memes, and much more.
(2) Secondly, the term ‘fake news’ is not only inadequate, but also misleading, because it has been
appropriated by some politicians and their supporters, who use the term to dismiss coverage
that they find disagreeable and has thus become a weapon with which powerful actors can
interfere in circulation of information and attack and undermine independent news media.”
(Publications.europa.eu. 2018).
Wardle and Derakhshan (2018), gave similar reasons as the EU expert group for their avoidance of the term
fake news on their report “Information Disorder”. Their first reason is, fake news is insufficient to describe
the complex phenomena of information pollution. Secondly, fake news is being used and abused by
politicians to attack news organizations and undermine press freedom. They argue that fake news lost its
academic meaning and understanding.
An 11 years, large-scale study on twitter (Vosoughi et al., 2018) also refrains from using fake news. The
authors believed that fake news is abused and misused by politicians beyond repair. As a reason of that it is
missing its understanding for academic purposes. Instead they preferred to use “true” and “false” news.
Even though they share the same concern, other scholars (Lazer et al., 2018) state three reasons for using
the term “fake news” on their articles.
First, it has a useful scientific meaning (the intersection of misinformation and mimicry of
traditional news media). Second, it is very prominent recent misuse has a salutary side effect of
focusing attention on the more general problem of misinformation. Third, while the term “false
9
news” avoids the weaponization problem in the near-term, should the effort to shift the popular and
scholarly nomenclature toward “false news” succeed, it would likely be subject to the same sort of
weaponization that we have seen with “fake news” (Lazer et al., 2018).
2.2 Fake news, post-truth and Gaudy Facts
Fake news has created an on-going debate on another front in relation to its representation of the situations
created by misinformation and disinformation. Like fake news, post-truth also declared word of the year by
Oxford |Dictionary (Flood, 2016).
McManus and Michaud, in their article “Never mind the buzzwords: defining fake news and post-truth”,
which is included on the “Fake news: A Road Map” (Haiden et al., 2018), define post-truth by explaining
what the prefix post means in linguistic terms. “Post suggests that the specified concept has become
unimportant or irrelevant. Post-truth would therefore imply that truth is no longer relevant, and more
importantly suggest that it was preceded at some point in time by an era of truth” (Haiden et al., 2018). Post-
truth politics on the other hand is defined as “where emotions are dominant the factual rebuttals or facts
checks are ignored on the basis that they are mere assertions” (Suiter 2016). The Oxford Dictionary defined
post-truth as “relating to or denoting circumstances in which objective facts are less influential in shaping
political debate or public opinion than appeals to emotion and personal belief” (Oxford Dictionaries 2018b).
Haiden (2018) argues that if post-truth examined in the context of populism, the meaning might be slightly
different. She stresses that “while truth is still important, instead of scientific research, personal experiences
and emotions are favored as the guiding principles for making the right judgements and seeking truth”.
Lewandowsky et al., (2017) state that ‘fake news’ and ‘post-truth’ are unknown to the public before 2016,
the year of the controversial US election. According to them “societal megatrends” are the main reasons for
the wider use and recognition of those two terms. The societal megatrends encompass “a decline in societal
capital, growing economic inequality, increased polarization, declining trust in science, and an increasingly
fractionalized media landscape”. Haiden (2018) has a similar view of post truth. She sees the year as the
start of the era of post-truth. “The year we allegedly left the world of rational argument and objective facts
and entered a world of ‘bullshit’ and lies”.
Some scholars argue that the situations that are happening since 2016 should only be expressed as post-truth
and they discard fake news as insufficient to express the events. Rochin (2017), discusses the
misunderstanding of fake news and what it represents. “There is a misunderstanding of what fake news can
be contested in an intellectual spectrum of true-untrue in an era of post-truth and mass social divide, this is
no longer viable”. Rochin also argues that fake news does not mean fact less or no more malicious news.
The reason is that fake news is being utilized to attack a person’s pre-existing belief. He explains this sad
demotion of the term as “the truth of the post-truth era”.
Rochin (2017) further explains why the term fake news should not be taken seriously in academia. He
believes that “fake news has become almost a joke; ‘a tongue in cheek reference’ used by society to refer to
any news it doesn’t agree with”. He also makes a distinction between fake news and true “fake news”. He
defined the real fake news as “a knowingly false headline and story written and published on a website that
is designed to look like a real news site and is spread via social media” and fake news is “Any story that
goes against one’s personal beliefs or feeling’s” (Rochin 2017).
10
There is also a strong opposition against the use of post-truth. The computer science professor Hal Berghel
(2017) strongly opposes the use of post-truth for representing the “the 2016 mood”. Berghel dismisses post-
truth and introduce his own term “gaudy facts”. He believes that his term will emphasize the garish and
tawdry nature of the recent political dialogue. He explains “gaudy facts have the advantage of avoiding the
word truth altogether since there’s precious little of that in political discourse”.
2.3 Scholarly definitions of fake news, disinformation, misinformation, propaganda and hoax
Several scholarly articles use the terms fake news, misinformation and disinformation interchangeably.
There is a growing concern about the lack of clear distinction between fake news, ideologically slanted
news, disinformation, misinformation and propaganda (Newman et al., 2017). Those terms have similarities
and differences in their definitions. In this section some of the definitions will be discussed.
The threats of fake news are worrying many including religious leaders. Pope Francis labeled fake news as
evil by stating the practice as old as humans. In his annual social communications message, the pope
explains fake news dissemination tactics as “Snake tactics” by refereeing the “Crafty serpent” from the book
of Genesis. According to the pope’s message the first fake news was created “the tragic history of sin”
(Francis, 2018).
Pope Francis used similar terminologies used by different scholars to define fake news and why and how it
spread like uncontrollable wildfire.
“False information based on non-existent or distorted data meant to deceive and manipulate the
reader” (Francis, 2018).
2.3.1 Defining fake news
Before entering the debate how algorithms define fake news, we have to get a better glimpse of how fake
news and its cousins are defined by academia. Fake news is recognized as the most ambiguous and misused
word. The definitions, explanations and understanding of what the term represents depends on the purpose
and aim of the definer. For this study, we tried to collect definitions, categorizations and classifications of
fake news from scholarly articles. The purpose of conducting a review of fake news definitions is to indicate
the need of an agreed upon meaning so that scholars, students and media professionals have a common
understanding. Mainly it helps those who are trying to combat fake news.
- Fake news, “News articles that are intentionally and verifiably false and could mislead readers”
(Allcott and Gentzkow 2017).
- “Fake news is a provocative headline that is shared and believed at face value, with no thoughtful
investigation” (Rochin 2017).
- Fake news represents information of various stripes that is presented as real but is patently false,
fabricated, or exaggerated to the point where it no longer corresponds to reality; what is more, this
11
information operates in the express interests of deceiving or misleading a targeted or imagined
audience (Reilly, 2018).
- “Fake news is the promotion and propagation of news articles via social media. These articles are
promoted in such a way that they appear to be spread by other users, as opposed to being paid-for
advertising. The news stories distributed are designed to influence or manipulate users’ opinions
on a certain topic towards certain objectives” (Gu et al., 2017).
- “Fake news defined as misinformation that has the trappings of traditional news media with
presumed associated editorial processes” (Lazar et al., 2017).
- Fake news refers to “lies represented as news that is, falsehoods online formatted and circulated
in such a way that a reader might mistake them for legitimate news articles” (Mustafariaj and
Metaxas, 2017).
- “Fake news, or hoax news, refers to intentionally false information or propaganda published under
the disguise of being authentic” (Svärd and Rumman, 2017).
- “The dissemination of false information via media channels (print, broadcast, online). This can be
deliberate (disinformation) but can also be the result of an honest mistake or negligence
(misinformation)” (Haiden et al., 2018).
- “Fabricated information that mimics news media content in form but not in organizational process
or intent” (Lazer et al., 2018).
- “Misleading news stories that come from non-reputable sources” (Gilda 2017).
- “Fake news is a recent popular and purposefully ambiguous term for false news stories that are
packaged and published as if they were genuine. The ambiguity of this term—an inherent property
of what it tries to label— makes its use attractive across the political spectrum, where any
information that conflicts with an ideology can be labeled ‘fake’” (DiFranzo and Gloria, 2017).
Based on the above 11 definitions we can conclude that fake news definitions are many and varied. False
news, misleading, fabricated, exaggeration, misinformation, online falsehood, intentional false information,
hoax and propaganda are some of the terms used to define fake news. Those terms are ambiguous and mixed.
The inclusion of misinformation for defining fake news can be taken as an example to illustrate the
confusion. Unlike fake news, misinformation represents mistakes or errors that happen as a result of
negligence. It does not have the intent to mislead or do harm.
2.3.2 Typology of Fake News
With all the complexity encircling the term fake news, some scholars focus on defining it while others
prioritize to classify fake news based on its use and its relationship with other information categories.
12
Wardle (2017) presents a typology of fake news which has seven categories:
Fig 1. Typology of misinformation and disinformation (Wardle, 2017)
Newman et al., (2017) conducted a survey for the Reuters digital news report (2017) in 36 countries. The
researchers were looking how fake news was understood and interpreted by different countries and cultures.
The result shows that the survey respondents failed at categorizing the three kinds of fake news separately
and got them mixed up. The following categories are identified by the researchers.
(1) news that is ‘invented’ to make money or discredit others;

(2) news that has a basis in fact, but is ‘spun’ to suit a particular agenda; and
(3) news that people don’t feel comfortable about or don’t agree with
A large-scale study by Tandoc et al. (2018), reviewed 34 scholarly articles which used the term “fake news”
between 2003-2017. They distinguished two dimensions: facticity and deception. Facticity was defined as
the degree to which fake news relies on facts. Deception was seen as the degree to which the creator of fake
news intends to mislead. Using these two dimensions the study resulted in “a typology of types of fake
news”. Six identified types of fake news:
(1) News satire- Referring to mock news programs, which typically use humor or exaggeration to present
audiences with news updates. These programs are typically focused on current affairs and often use the
style of a television news broadcast (a “talking head” behind a desk, with illustrative graphics and
video), much as a regular news program.
(2) News parody- Parody is a second format which previous studies have referred to as fake news. It shares
many characteristics with satire as both rely on humor as a means of drawing an audience. It also uses
a presentation format which mimics mainstream news media. Where parodies differ from satires is their
use of non-factual information to inject humor. Instead of providing direct commentary on current
affairs through humor, parody plays on the ludicrousness of issues and highlights them by making up
entirely fictitious news stories.
(3) Fabrication- refers to articles which have no factual basis but are published in the style of news articles
to create legitimacy. Unlike parody, there is no implicit understanding between the author and the
13
reader that the item is false. Indeed, the intention is often quite the opposite. The producer of the item
often has the intention of misinforming.
(4) Photo Manipulation- the manipulation of real images or videos to create a false narrative. Where the
previous categories generally refer to text-based items, this category describes visual news.
Manipulation of images has become an increasingly common occurrence with the advent of digital
photos, powerful image manipulation software, and knowledge of techniques. Effects may range from
simple to complex. Simple adjustments can include increasing color saturation and removing minor
elements. More-invasive changes can include removing or inserting a person into an image.
(5) Advertising and Public Relations- Advertising materials in the guise of genuine news reports as well
as to refer to press releases published as news. based on this definition video news releases are termed
as fake news. VNRs are pre-packaged video segments produced by public relations firms aimed at
selling or promoting a product, a company, or an idea.
(6) Propaganda- Refers to news stories which are created by a political entity to influence public
perceptions. The overt purpose is to benefit a public figure, organization or government. (Tandoc et al.,
2017).
2.3.3 Disinformation
Since the era of the cold war, the term disinformation has been a discussion point both in the media and
academia. Martin (1982), claims that disinformation is derived from the Soviet word ‘Dezinformatsiya’
which is described as the dissemination of false and provocative information. “The Soviet Union secret
service used it as a persuasive technique that’s based on forgeries and staged events”. He classified forgery
and fabrication as parts of disinformation.
Bittman (1990), argues that the term disinformation has its roots from German and adopted by the former
Soviets to deceive their ideological opponents. Both Martin and Bittman classify disinformation as a form
of propaganda.
- “False, inaccurate, or misleading information designed, presented and promoted to intentionally

cause public harm or for profit” Publications.europa.eu. (2018).
- “Disinformation refers to the deliberate creation and sharing of information known to be false”
(Wardle,2017).
- “Is a form of propaganda in which either the focus is based on some unlawful act or the message is
a misrepresentation of a lawful act or true situation. The intent of disinformation is to persuade by
whatever means possible or available” (Martin, 1982).
14
- “Disinformation is twisted. Secretly inserted into an opponent’s communication system, its intention
is to deceive either the public in which case its propagandistic disinformation or the decision-
making elite” (Bittman, 1990).
- “Disinformation is intentionally false or inaccurate information that is spread deliberately” (Kragh

and Åsberg, 2017).
- “Disinformation is nothing new, of course. Forged documents, doctored photographs, deceptive

advertising, deliberately falsified maps, and government propaganda” (Fallis, 2009).
- Dis-information- “Information that is false and deliberately created to harm a person,

social group, organization or country” (Wardle and Derakhshan, 2018).
The reviewed seven definitions and related terms are used to define disinformation. False, inaccurate,
misleading, deliberate creation, a form of propaganda, forgery, and deliberately created have more or less
an identical meaning. The definitions seem to follow the intent behind disinformation causing public harm,
gaining financial benefit, to persuade, to deceive and to cause harm to the public and country are stated as
intents to disinform.
2.3.4 Misinformation
Several academic articles define misinformation as an error, honest mistake and inaccurate information. It
doesn’t have the intent to deceive but it is misleading. (Fallis, 2015). Wardle and Derakhshan (2018),
categorized three types of information disorder. In addition to disinformation and misinformation they
include mal-information, it is defined as “Information that is based on reality, used to inflict harm on a
person, organization or country” (Wardle and Derakhshan, 2018, p.21).
- “Is wrong or false information circulated as a result of a genuine mistake, omission, prejudice or
sheer ignorance” (Bittman 1990).
- “Misinformation is unintentional dissemination of false information” (Kragh and Åsberg, 2017).
- “Misinformation refers to the inadvertent sharing of false information” (Wardle, 2017).
- “Misinformation is inaccurate information that is the result of an honest mistake or of negligence”

(Haiden et al., 2018).
- “Misleading or inaccurate information shared by people who do not recognize it as such”

Publications.europa.eu. (2018).
The above five have a coherent definition of misinformation. It seems that the academia has a common and
less ambiguous understanding of the definitions. Several papers and articles defined misinformation in the
same fashion and we exclude them to avoid repetition.
15
2.3.5 Propaganda
Bittman (1990), and Martin (1982), explain that disinformation is a crucial component of propaganda. The
commonality between fake news, disinformation and propaganda is the fabrication of and the intent to
deceive. Martin also states that propaganda is a failed attempt of fabrication and falsification for the purpose
of persuasion and deception.
- “Propaganda is something that is conducted by one's opponents, and it must be unsuccessful to be

called propaganda” (Martin,1982).
2.3.6 Hoax
Hoax has some similarities with disinformation. The intent is to deceive by the deliberate fabrication of a
falsehood. The Oxford Dictionary defines hoax as a humorous or malicious deception, and trick or deceive
someone. Wikipedia states several assumptions and perceptions of hoax. As propaganda, hoax also used
from individuals to governments to deceive public opinion. A hoax is often intended as a practical joke or
to cause embarrassment, or to provoke social or political change by raising people's awareness of something
(Wikipedia).
- “To trick into believing or accepting as genuine something false and often preposterous”
(Merriam-webster.com).
- Hoaxing as “a humorous or mischievous deception, usually taking the form of a fabrication of

something fictitious or erroneous.” (Concise Oxford dictionary).
- “A hoax is a falsehood deliberately fabricated to masquerade as the truth” (Wikipedia.com).
The above listed definitions show that defining information disorders is not an easy task. Fake news has the
most diverse definitions. The intent to deceive and to cause harm can be taken as the distinguishing factor
that disinformation, propaganda, and hoax share. We will use those definitions as a stepping stone for
examining the definitions given by algorithmic deception detection solutions.
16
3 Theory
The thesis focuses on the definitions of fake news, disinformation and misinformation for algorithmic
detection purposes. How words, terms and sentences are being defined has a big impact on the detection
process. In this part of the study we discuss the theory of meaning, especially semantics and pragmatics
meaning, and briefly the concept of deception detection.
3.1 Meaning
Words, terms, utterances, and sentences can have several meanings based on the context and the intent of
the speaker. Words can be used to mean something else than the word reads and sounds. Words have
meanings that have ‘implicature’ (Grice 1967). If the words are expressed with identifiers like “means”,
“means something”, “means that”, we can say the word retains its first or natural sense (Grice 1957). This
can be termed as the natural sense or simply sense. Grice termed the second sense of the word as non-natural
sense. He denotes this unnatural sense as “means NN”.
Other scholars (Larson and Segal 1995; Thomas 1994, 2014; Hofmann 2015) have examined the ambiguity
created by contextual meanings using the theories of semantics and pragmatics. Semantics represents the
envisioned meaning of the word. The thesis will focus on semantics to show the difficulty of defining some
terms like fake news for classifying and categorizing false and misleading news stories. First, semantics will
be discussed for situating pragmatics. Pragmatics will be discussed thereafter.
3.1.1 Semantics
Semantics is the study of linguistic meaning. The aim is to provide theoretical descriptions and explanations
of the phenomena of linguistic meaning. Larson and Segal (1995, p.2) identified actual meaning, ambiguity
and anomaly as semantic properties.
Actual meaning is the intended meaning of a word without any context attached. Larson and Segal (1995
p.2), explained actual meaning with an example; “Camels have humps means Camels have humps”. Actual
meanings should be taken literally.
Ambiguity can arise from one or more components of a sentence. They presented at least three types of
ambiguity;
(a) Ambiguity that arises from one of the component words.
(b) Ambiguity arises not from the words of the clause but rather from our understanding of those
words in combination.
(c) Ambiguity involves the combination of (a) and (b).
The third semantics property is anomaly. It arises when the property has an aberrant meaning. The authors
borrowed Chomsky’s (1957) “colorless green ideas sleep furiously” to explain the property. The quoted
sentence can be identified as an odd sentence with some absurdity. Unusual but syntactically correct
sentences are termed as parts of semantics.
According to Larson and Segal, the process of using knowledge of language in understanding perceived
sentences involves three processes of understanding.
17
(1) Parsing - is understanding the sentences in one’s language phonologically. Syntactic arrangement
must be identified, and one must know the meaning of the words and how they compose. The
concept seems clear and easy, but it should be done sequentially in order of phonological
(systematic organization of sounds in a language), syntax and then semantics. Parsing only concerns
the application of linguistic knowledge. Linguistic knowledge only provides us with context-
independent meaning of an utterance.
(2) Language understanding - contains context-dependent features of utterance meaning. relevant

features of context must be identified and combined with knowledge of language in (1).
(3) Pragmatics - knowledge of language only provide us with literal meanings of sentences. There is
always a gap between what is being said with a sentence and what the speaker intends to convey.
Pragmatics can bridge this gap.
3.1.2 Pragmatics
Pragmatics is the study of meaning that deals with words or sentences with more than one meaning.
Hofmann (2015) elaborates the role of pragmatics in understanding the intent of words with the following
example.
“Whenever language is used, there is a speaker and his intent, and more often than not, the ultimate
intent is hidden behind the literal meaning (i.e. 'between the lines') of what is said. A teacher might
say to a twenty-minute-late student, 'My, you're early today!' Does she mean that? What does she
mean?” (Hofmann 2015, p.273).
Hofmann states that pragmatics appears in everyday life. “in non-literal use of sentences, the idea conveyed
is not the same as the meanings of the word. Exaggeration is one very common type, as are metaphor,
sarcasm and irony” (Hofmann, 2015). He also mentions politeness as a reason why people avoid speaking
directly. Thomas (2014) also mentions politeness as a reason for indirectness. Politeness theory is seen as a
sub-discipline of pragmatics. Politeness doesn’t imply the speaker’s behavior. It deals with what the speaker
says and how the hearer reacts (Thomas 2014, p.150). Conversational maxims are one of the theories
proposed to understand the non-literal use of the word. The theory is developed to secure understanding
between the speaker and the listener. The two parties must have some level of understanding and one must
‘choose one’s words’ so that the other can understand the intent and the hearer must try to figure out what
the speaker meant. Hofmann proposes six of such maxims;
Maxims of quantity
(1) Give as much information as is needed.
(2) Give no more information than is needed.
Maxims of quality
(3) Do not say what you believe to be false.
(4) Do not say what you have no evidence for.
Other maxims
18
(5) Be relevant.
(6) Be perspicuous:
(a) do not use obscure expressions.
(b) do not use ambiguous expressions unless necessary.
(c) be brief.
(d) be orderly (Hofmann 2015 p.275).
The most common definition of pragmatics is meaning in use or meaning in context. Thomas (2014)
introduces pragmatics with the following example;
“I might say: “It's hot in here!”, but what I mean is: “Please open the window!” or “Is it all right
if I open the window?” or “You're wasting electricity!” People can mean something quite different
from what their words say, or even just the opposite” (Thomas, 2014 p.1)
Thomas prefers to define pragmatics by compartmentalizing it into levels of meaning. abstract meaning,
contextual meaning and force.
Abstract meaning - is concerned with what a word, phrase, sentence, etc. could mean. A dictionary
meaning of words or phrases can be a good example.
Contextual meaning (utterance meaning or 1st level of speaker meaning) - is the transition we as a hearer
make when we understand what the speaker means by those words on that particular occasion. In the above
example, when the speaker utters the words “It’s hot in here!” the abstract meaning is that the room is
obviously hot. When the hearer/listener acts after understanding the intent of the words, it can be said that
he/she arrived at the contextual meaning (Thomas, 2014).
Force - refers to the speaker’s communicative intentions. Thomas describes force as the second level of
speaker meaning. Force deals with finding the real intention behind the word, utterance or question. Thomas
explains force with the question “Is it your car?”. Here we can’t find any ambiguity or hidden meaning. It
is a clean and simple question. In this regard the question passes both abstract and contextual phases easily.
Here the point of concern is ‘the force of the question’ (Thomas 2014). Why did he ask? Is he appreciating
or expressing scorn? Is he commenting how the driver parked his car? Is he asking for a ride? This and other
‘pragmatics forces’ (Thomas 2014) can be drawn from a simple question or a sentence when a listener tries
to understand the meaning.
Pragmatics and indirectness
Indirectness is the major manifestation of pragmatics. Thomas raises four points we must consider when we
discuss pragmatic indirectness.
- We shall be concerned with intentional indirectness.
- Indirectness is costly and risky.
- We assume (unless we have evidence to the contrary) that the speakers are behaving in a rational
manner and, given the universality of indirectness, that they obtain some social or communicative
advantage through employing indirectness.
- In discussions we might tend to ignore some terms or ideas as difficult to express and we prefer to
avoid them. Thomas (2014) argues that for the purposes of any particular argument, we shall ignore
the possibility that X cannot be expressed.
19
3.2 Deception Detection
Deception is often used interchangeably with lie. It is defined as intentionally, knowingly, and/or purposely
misleading another person (Levine, 2014). Buller and Burgoon (1994), define deception as messages and
information knowingly transmitted to create a false conclusion. Levine (2014) argues that deception does
not require ‘conscious forethought’. But he also admits that some deception needs planning. Deceptive
messages involve intent, awareness, and/or purpose to mislead (Levine, 2014). He argues that deception is
probable when the truth is difficult to bear or when it becomes inefficient for the intended purpose of the
deceiver. Based on the above definitions and descriptions we can assume that deception detection is a
question of pragmatics. Sille, (2018) expresses his concern on the efficiency of algorithms to detect
information disorders without realizing the pragmatic meaning of the word, utterance and term.
Zhou et al., (2004), proposed an automated linguistics-based cues (LBC) methods for deception detection.
The automatic deception detection could be based on machine learning techniques. Their theory involves
training a machine learning algorithm with previously classified messages in each context. After the
algorithms had been successfully trained the resulting cues or indicators could be used to create a set of
profiles for deceptive messages in that context. The values of indicators could be fed as features to an
algorithm that learns to combine evidence to generate high-confidence warnings of deception (Zhou et al.,
2004). The researchers are confident that any algorithm trained by their method will have the ability to
adjust to different strategies of deception appearing in different contexts.
Zhou et al. (2004) concluded that based on their findings, automated LBC is achievable. But they have
doubts that the same cue profiles are unlikely to apply uniformly across contexts.
3.3 Related Work
Fake news dissemination on social media got a wider attention in the academic world since the 2016 US
election. Several researches are conducted to indicate ways to combat fake news. Researchers are
developing theories, theoretical frame works, algorithmic and other information systems and computer
science-based solutions. we tried searching for scholarly articles on the concern we raised, and we only
found two articles on the specific area.
Sille Obelitz Søe’s conceptual paper “Algorithmic detection of misinformation and disinformation: Gricean
perspectives” published on Emerald in 2018.
With the outset of automatic detection of information, misinformation, and disinformation, Sille’s paper has
the purpose of examining and discussing various conceptions of information, misinformation, and
disinformation within philosophy of information. The examinations are conducted within a Gricean
framework to account for the communicative aspects of information, misinformation, and disinformation as
well as the detection enterprise.
While there often is an exclusive focus on truth and falsity as that which distinguish information from
misinformation and disinformation, the paper finds that the distinguishing features are
intention/intentionality and non-misleadingness/misleadingness with non-misleadingness/misleadingness
20
as the primary feature. Further, the paper rehearses the argument in favor of a true variety of disinformation
and extends this argument to include true misinformation.
Sille stresses that to determine whether a news is misinformation or disinformation it requires evaluative
judgements of content, context, purpose etc. and whether such judgments can be automated. One of his main
questions is what algorithms should look for to detect misinformation and disinformation. He investigated
four algorithmic detection solutions and studied how they defined misinformation and disinformation for
detection purposes. The conceptual paper explores and analyzes the very nature of information,
misinformation and disinformation and their interactions. Sille proposes a more detailed study is required
for better understanding of those notions. He argues that algorithmic detection solutions should detect the
distinguishing features of those notions.
Haiden and Althuis (2018) paper “The definitional challenges of fake news” focuses on investigating how
fake news and disinformation defined and how it impacted research on the field. The authors reviewed most
cited papers published since 2016. They evaluated the definitions they gather against the definitions
published on a scholarly paper titled, “Fake News: A Road Map” (2018). They didn’t state their reason for
specifically selecting the paper for evaluation purposes except its being a comprehensive study. The authors
main aim is to bridge the gap on defining fake news, disinformation and misinformation between policy
makers and the research community. The paper finds that policy makers are more focused on the definitions
of disinformation while neglecting misinformation. The paper acknowledges the efforts done by algorithms,
but it didn’t discuss how fake news and disinformation are defined for algorithmic detection.
21
4 Methodology
4.1 Research Strategy
The study aimed at the meaning, understanding and interpretation of fake news, misinformation,
disinformation and hoax focusing on algorithm-based deception detection scientific papers. The study has
at least two ‘hypotheses’ to prove;
A) Fake news is badly or inadequately defined.

B) Deception detection algorithms define fake news insufficiently.
The thesis follows qualitative approach for data collection and thematic analysis to analyze the data. Since
fake news is not clearly and acceptably defined, exploratory approach chosen to generate qualitative
information and interpretation. Meyers (1997), defines qualitative research as a method that involves the
use of qualitative data, such as interviews, documents, and participant observation data to understand and
explain social phenomena. The thesis is an exploratory study with the aim of understanding the fake news
phenomena through the meanings and interpretation provided by researchers. “Exploratory research is
intended to provide conclusive, but it helps to get a better understanding of the problem” Research-
Methodology (2018).
4.1.1 Mapping Definitions

The results of the 2016 US election and the Brexit referendum sent a warning signal to other nations that no
country is safe if the superpowers fail for it. Political scientists, journalists, information professionals,
educators and the tech community were working separately and seldom in coordination to combat fake
news. We searched most of the peer-reviewed scholarly articles published on journals, conferences and on
library databases. We mapped the selected articles in tabular format. The categories we identified are; Title,
Authors, publisher, year of publication, method of detection, artifact/model/framework, and definition of
fake news. The mapping exercise showed that the definitions and interpretations presented by those articles
are not in unison, there is no agreed upon definition of fake news.
Some of the definitions (discussed on section 2.3) are presented to show how fake news, misinformation,
disinformation, hoax and propaganda are defined differently. Fake news detection is a new field of study
with limited articles on the subject. Most of the algorithmic solutions are based on machine learning
principles.
Structure of the mapping

The algorithmic deception detection research articles are searched mostly on Google Scholar. Other
scientific journals, conferences and university library online data bases are also searched. The main criteria
to pass an article as valid was based on; 1) The article need to be algorithmic deception detection solution.
2) The article should define at least one information disorder. 3) Peer reviewed, if applicable.
The following terms are used separately and in conjunction with other listed terms.
- “fake news”
22
- “misinformation”
- “disinformation”
- “fake news detection”
- “deception detection”
- “misinformation detection”
- “disinformation detection”
- “hoax detection”
- “algorithmic deception detection”
- “automatic deception detection”
Most of those key words yield the desired deception detection algorithmic articles. “deception detection”,
and “algorithmic deception detection” produced mixed results from other branches of detection. “fake
news”, “misinformation” and “disinformation” delivered very few articles. The rest of the key words
repeatedly shown similar results. 27 articles were collected from;
- IEEE publications,
- Arxiv.org,
- willy.com,
- ACM digital library,
- research gate
- EMNLP workshop,
- ACL web,
- Semantic scholar,
- and others.
Based on the selection criteria,16 articles were selected. We only selected those that are defined at least one
of the information disorders. Out of the 27 articles 4 of them failed to define what they are proposing to
detect. As a consequence, they were excluded from the study. Five articles were omitted because they did
not deal with technical algorithmic solutions; some of them are frameworks or a study about detection
methods.
This is the list of 16 scholarly articles selected. Only the basic information of the publications is stated in
the table. The definitions will be discussed in the results section in Chapter 5.
Id Year Publication details
1 2015 Title: Automatic Deception detection: Methods for finding fake news
Author(s): Niall J. Conroy, Victoria L. Rubin, and Yimin Chen
Summary: The research surveyed the technologies developed for the
purpose of deception detection. It researched different veracity
assessment method developing from linguistics cue approaches with
machine learning and network analysis approaches. As a result, they
proposed operational guidelines for a feasible fake news detecting
system.
Keywords: Deception detection, fake news detection, veracity
assessment, news verification, methods, automation, SVM,
knowledge networks, predictive modelling, fraud
2 2015 Title: Deception detection for news: Three types of fakes
23
Author(s): Victoria L. Rubin, Yimin Chen and Niall J. Conroy
Summary: The paper discusses the three types of fake news identified
by the authors for text analysis and predictive modeling corpus. The
study proposes guidelines.
Keywords: news verification; deception detection; fake news
detection; credibility assessment; reputable sources; fabrication; hoax;
satire; natural language processing; text analytics; predictive modeling;
corpus construction.
3 2017 Title: Some like it hoax: Automatic fake news detection on social
networks
Author(s): Eugenio Tacchini, Gabriele Ballarin, Marco L. Della
Vedova, Stefano Moret, and Luca de Alfaro
Summary: the researchers developed a system that can predict
Facebook posts as hoaxes and non-hoaxes based on the users who
‘liked’ them. The study uses two classification methods based on
logistic regression and on a Boolean crowdsourcing algorithm.
Keywords:
4 2017 Title: CSI: Hybrid deep model for fake news detection
Author(s): Natali Ruchansky, Sungyong Seo, Yan Liu
Summary: the study proposed a model, CSI (capture, score and
integrate) that compasses the three characteristics of fake news; the text
of the article, the user response it receives and the source the user
promoting it. Recurrent neural network is used to capture the temporal
pattern of user activity on a given article.
Keywords: Fake news detection, Neural networks, Deep learning,
Social networks, Group anomaly detection, Temporal analysis.
5 2017 Title: Fake news detection on social media.
Author(s): Kai Shu, Amy Sliva, Suhang Wang, Jiliang Tang, and Huan
Liu
Summary: In this survey, it presented a comprehensive review of
detecting fake news on social media, including fake news
characterizations on psychology and social theories, existing algorithms
from a data mining perspective, evaluation metrics and representative
datasets.
Keywords:
6 2018 Title: Detecting opinion spams & Fake News using text classification
Author(s): Hadeer Ahmed, Issa Traore, and Sherif Saad
Summary: the researchers Introduced a new n-gram model to detect
automatically fake contents with a particular focus on fake reviews
and fake news by comparing 2 different features of extraction
techniques and 6 machine learning classification techniques.
Keywords: fake content detection, online fake news, online fake
reviews, online social network security, opinion spams, text
classification
7 2015 Title: Towards news verification: deception detection methods for
news discourse
24
Author(s): Victoria Rubin and Yimin Chen
Summary: the study focuses on automated news verification
techniques. Vector space model used to cluster news by discourse
feature similarity.
Keywords:
8 2016 Title: Hoaxy: A traction for tracking online misinformation
Author(s): Chengcheng Shao, Giovanni Luca Ciampaglia,
Alessandro Flammini, and Filippo Menczer
Summary: The researchers developed Hoaxy, a platform for the
collection, detection, and analysis of online misinformation and its
related factchecking efforts by using preliminary analysis of a sample
of public tweets containing both fake news and fact checking.
Keywords: Misinformation; hoaxes; fake news; rumor tracking; fact
checking; Twitter
9 2017 Title: Automatic detection of fake news
Author(s): Veronica Perez-Rosas, Bennett Kleinberg, Alexandra
Lefevre, and Rada Mihalcea
Summary: the study focused on automatic identification of
fake content in online news. they introduced two novel datasets for the
task of fake news detection, covering seven different news domains
secondly, they conduct a set of learning experiments to build accurate
fake news detectors.
Keywords:
10 2018 Title: Fake news detection in social network via crowd signals
Author(s): Sebastian Tschiatschek, Adish Singla, Manuel Gomez
Rodriguez, Arpit Merchant, and Andreas Krause
Summary: the researchers used leveraging crowd signals for
detecting fake news and is motivated by tools recently introduced by
Facebook that enable users to flag fake news. they proposed a novel
algorithm, detective, that performs Bayesian inference for detecting
fake news and jointly learns about users’ flagging accuracy over
time.
Keywords:
11 2017 Title: Fake News Detection using Stacked Ensemble of Classifiers
Author(s): James Thorne, Mingjie Chen, Giorgos Myrianthous,
Jiashu Pu, Xiaoxuan Wang, and Andreas Vlachos
Summary: stacking ensemble method for the purpose and obtained
improvements in classification accuracy exceeding each of the
individual models’ performance on the development data.
Keywords:
12 2017 Title: Stance Detection for the Fake News Challenge with Attention
and Conditional Encoding
Author(s): Stephen Pfohl, Oskar Triebe, and Ferdinand Legros
Summary: By applying the concepts of neural attention and
conditional encoding to long short-term memory networks (LSTM)
the researchers developed a stance detection tool.
Keywords:
13 2018 Title: We Built a Fake News & Click-bait Filter: What Happened
Next Will Blow Your Mind!
Author(s): Georgi Karadzhov, Pepa Gencheva, Preslav Nakov, and
Ivan Koychev
25
Summary: the researchers utilized machine learning features like
neural networks, attention mechanism, sentiment lexicons, author
profiling, Lexical features, semantic features to develop a fake news
filter.
Keywords:
14 2016 Title: Disinformation on the web: Impact, Characteristics, and
Detection of Wikipedia Hoaxes
Author(s): Srijan Kumar, Robert West, Jure Leskovec
Summary: the researchers study an automated classifier of hoaxes for
the purpose of detection on Wikipedia.
Keywords:
15 2018 Title: Studying Fake News via Network Analysis: Detection and
Mitigation
Author(s): Kai Shu, H. Russell Bernard, and Huan Liu
Summary: The research is aimed at reviewing network properties
for studying fake news, introduce popular network types and propose
how these networks can be used to detect and mitigate fake news on
social media.
Keywords: Fake news, network analysis, social media
16 2010 Title: On deception detection in multi agent systems
Author(s): Eugene Santos, Jr., and Deqing Li
Summary: The study detect deception by observing the correlations
between agents, which can be used to make a reasonable prediction of
the agents ‘reasoning processes.
Keywords: Bayesian networks (BNs), deception detection,
multiagent system, parametric study.
- Table 1 publication details
4.2 Coding and analysis of definitions

Thematic analysis is one of the methods proposed for successful analysis of qualitative research. We
selected thematic analysis to perform the coding and analysis on the mapped definitions. Braun and Clarke
(2014), defined thematic analysis as a method for systematically identifying, organizing and offering insight
into patterns of meaning (themes) across a data set. Thematic analysis is trusted with the ability to let
researchers see and make sense of collective or shared meanings and experiences (Braun and Clarke, 2014).
The patterns researchers collected by thematic analysis needs to be important regarding a specific topic or
the research question. The other reason for researchers to select thematic analysis is its flexibility to let the
researchers see the data in different ways (Braun and Clarke, 2014).
The idea behind the thematic analysis is to get a firmer or better definition of fake news or information
disorder by looking at how well the detection algorithms define and recognize the reality of fake news or
information disorder. The study will look into patterns, shared meanings and perspective of fake news
definitions in the 16 articles. The study will assess the definitions at word level. All the terms and words
will be examined if they are representing the information disorder they are used to define. By looking into
shared meanings, we will try to indicate the information disorder the words and terms should represent or
define.
26
4.2.1 Methods of Thematic analysis
Inductive vs Deductive
Themes or patterns of data can be identified by induction, deduction or by a mix method. Induction is a
‘bottom up’ way. The themes are strongly linked to the data, this means the data are specifically collected
for the research in terms of interview or document. It is a data driven approach. (Braun and Clarke, 2006).
Deduction on the other hand is based on the researcher’s theoretical or analytical interest and it is an analyst
driven approach. This study is theoretical, it has a research question to get answers for. The articles are
selected for the purpose of examining their definitions and to determine whether the themes we are looking
for are linked to the data.
Semantic vs Latent
Thematic analysis uses semantic and latent approaches for identifying themes. In Semantic, the themes are
identified within the explicit or surface meanings of the data and the analyst is not looking for anything
beyond what a participant has said or what has been written (Braun and Clarke, 2006). A latent or pragmatic
approach detects contextual themes. “the latent level goes beyond the semantic content of the data, and starts
to identify or examine the underlying ideas, assumptions, and conceptualizations – and ideologies - that are
theorized as shaping or informing the semantic content of the data” (Braun and Clarke, 2006). The thesis
discussed (on section 3.1) semantics and pragmatics meanings in relation to defining words, terms and
sentences. Since the thesis is concerned with plain definitions without context, we selected semantic
approach.
What is considered as a theme?
The most central question in thematic analysis is what can be labeled as a theme or pattern. Based on the
argument of Braun and Clarke (2006), any part of the data that is labeled as important by the analyst can be
a theme. Braun and Clarke argue that it is all about prevalence both in time and space for each data item.
There is no clear-cut answer to the question of what proportion of the data set needs to display evidence of
the theme to be considered as a theme (Braun and Clarke, 2006).
4.2.2 Doing the analysis
Our thematic analysis follows the six phases that Braun and Clarke (2006) considered essential in their
thematic analysis framework. The phases are not linear so that the analyst can move up and down several
times between them.
Phase 1: knowing your data, Phase 4: Reviewing themes,

Phase 2: Generate initial codes, Phase 5: Define themes,
Phase 3: Searching for themes, Phase 6: produce the report.
Table 2: Six phase thematic analysis frame work. (Braun and Clarke, 2006).
27
Phase 1: Knowing your data
The selected data was read repeatedly. It was organized in tabular format for ease of access. The selected definitions
were repeatedly checked against the sense of text and utmost care and attention was given to acquire the definitions as
presented in the papers.
Phase 2: Generate initial codes
Next the initial codes were generated from the data gathered in phase 1.
Article Data items Initial codes

1 Fake news- is intentionally deceptive news report Fake news
Intentionally deceptive
2 Three types of fake news; Fake news
1) serious fabrication, Fraudulent reporting, falsification, Fabrication
or exaggeration uncovered in mainstream media or Fraudulent report
participant media, yellow press or tabloids. Falsification
Exaggeration
Mainstream media
Yellow press
Tabloids
Hoaxing
2) Hoaxing is another type of deliberate fabrication or Deliberate fabrication
falsification in the mainstream or social media. Fabrication
Falsification
Mainstream media
Social media
Mainstream journalism
3) Humorous fakes “in a format typical of mainstream Irony
journalism but rely heavily on irony and deadpan humor to Deadpan humor
emulate a genuine news source, mimicking credible news sources Mimicking
and stories, and often achieving wide distribution (news satire, Parody
parody, game shows) Game shows
3 Hoaxes- i.e. intentionally crafted fake information Hoax
Intentionally crafted
Fake information
4 Fake news is news content that is fabricated Fake news
Fabricated news content
5 Fake news- low quality news with intentionally false information Fake news
Low quality news
Intentionally false
information
6 Three fake news groups; Fake news
False news- which is news that is completely fake and is made up False news
by the writers of the articles. Completely fake
Made up
28
Fake satire
Fake satire news- which is fake news whose main purpose is to Fake news
provide humor to the readers. Humor
Poorly written
Not entirely accurate
Poorly written news article- which have a certain degree of real Fake news
news but are not entirely Misleading
accurate. Opinion spam
fake news and misleading articles are another type of opinion
spam.
7 Fake news-Fake, fabricated, falsified, disingenuous, or misleading Fake news
news reports constitute instances of digital deception or deliberate Fake
misinformation. Fabricated
Falsified
Disingenuous
Digital deception
Deliberate
Misinformation
Digital deception
Digital deception-intentional control of information in a Intentional
technologically mediated environment to create a false belief or control of information
false conclusion. False belief
False confusion
8 Misinformation- unintentional spread of false or inaccurate Misinformation
information. misinformation are rumors, hoaxes, fake news, and Unintentional
conspiracy theories stated as examples of misinformation. False
Inaccurate
Misinformation
Rumors
Hoaxes
Fake news
Conspiracy theories
9 Fake news- serious fabrications i.e. news items about false and Fake news
nonexistent events or information such as celebrity gossip Serious fabrication
False
Nonexistent
Celebrity gossip
10 Fake news also known as hoax, rumor etc. Fake news
Hoax
Rumor
11 Fake news-highly partisan fabricated materials on social media Fake news
Highly partisan
Fabricated material
Social media
29
12 Fake news- a made up story with an intention to deceive Fake news
Made up story
Intention to deceive
13 Fake news is written and published with the intent to mislead to Fake news
gain financially or politically, often targeting specific user groups.
Intent to mislead
click-baits, which are distinguished by their sensational, Sensational
exaggerated, or deliberately false headlines that grab attention and Exaggerated
deceive the user into clicking an article with questionable content. Deliberately false
headlines
Deceive
14 Misinformation- is conveyed in the honest but mistaken belief that Misinformation
the relayed incorrect facts are true. Honest mistake
Misinformation- information that is unintentionally false. Incorrect facts
Unintentionally false
Disinformation- false facts that are conceived to deliberately Disinformation
deceive or betray an audience. False facts
Deliberately deceive
Hoax
Hoax is a kind of disinformation. Disinformation
Hoax- a deliberately fabricated falsehood made to masquerade as Deliberately fabricated
truth. Falsehood
15 Fake news is news with intentionally false information Fake news
Intentional
False information
16 Deception as information designed to “manipulate the behavior of Deception
others by inducing them to accept a false or distorted presentation Manipulative
of their environment—physical, social, or political” and False
Deception as a “deliberate act perpetrated by a sender to engender Distorted
in a receiver’s beliefs contrary to what the sender believes is true Deliberate
to put the receiver at a disadvantage.” Misinformation
Mistake
Misinformation is defined as mistakenly providing the wrong Wrong information
information.
Table 3: Initial codes
Phase 3: Searching for themes

As shown in section 2.3.3 Fake news, hoax, rumor found in some definitions represent different kinds of
information disorder. Using those terms beyond their intended meaning causes confusion. Therefore, in the
third analysis phase, overarching themes and sub-themes were identified based on the initial codes.
30
Intentionally deceptive Fabrication
Falsification
False and non-existent
Made up
Misleading
False news
Hoax Deliberate fabrication
Falsification
Intentionally crafted
Fake information
Disinformation Deliberately deceive
False facts
Digital deception Intentional control of information
False belief
Manipulative
Distorted
Deliberate
Misinformation False information
Honest mistake
Incorrect facts
Wrong information
Rumor
Fake
Table 4: Searching for themes
Phase 4: Reviewing themes
Phase 4, according to Braun and Clarke (2006), needs two levels of reviewing and refining. First a review
at the level of the coded data, Second, a review at the level of the themes. On this phase we can rework the
themes if they seem unfit or create new theme, merge or discard. The selected themes and sub-themes are
refined. Some of the themes are collapsed and some merged with other themes and became sub-themes.
Because digital deception, disinformation, hoax and intentionally deceptive themes have similar sub-themes
and representations, they are merged under disinformation.
Disinformation Deliberate fabrication

Falsification
Intentionally false
Digital deception
Misleading
Misinformation Honest mistake
Wrong information
Incorrect facts
Misleading
Digital deception Manipulative
Distorted
Deliberate
Table 5: Reviewed themes
31
Phase 5: Defining themes
This is the final step to check if the thematic map helps to capture the essence of each theme and what
aspects of the data each theme captures (Braun and Clarke, 2006). The narrative of the theme should be in
line with the narrative of the data. In our research, it appeared that the searched themes delivered more
themes and sub-themes but at phase 5 it became clear that most of the sub-themes represented similar
concepts and definitions. Out of the 16 algorithmic deception detection papers, only 1 paper mentioned and
discussed disinformation. According to the studies discussed (on section 2.3) disinformation represents all
the information disorders mentioned in the 16 articles. The features of disinformation are found scattered
under fake news, hoax and digital deception. Disinformation and misinformation are identified as themes.
Even though misinformation agreeably defined as an honest mistake and without intention to deceive, it is
still as misleading as disinformation.
Falsification Deliberate Honest Wrong

Fabrication Mistake Information
Disinformation Misinformation
Intentionally Digital Incorrect

false deception facts
Misleading
Fig 2: Final Thematic Map
32
5 Results
In this chapter the themes and sub-themes identified in the final thematic map presented and discussed. To
evaluate the correctness of theme identification we tried to compare, contrast and relate the themes with the
definitions discussed in Chapter 2. The results shown that disinformation is the more appropriate term for
expressing and explaining the so called ‘fake news’ phenomenon for the purpose of algorithmic deception
detection.
5.1 Identified Themes

Based on the final thematic map disinformation (section 2.3.3) and misinformation (section 2.3.4) were
qualified as themes. The sub-themes; deliberate fabrication, intentionally false, digital deception,
falsification, misleading, honest mistake, incorrect facts, and wrong information assists to identify
disinformation and misinformation as information disorder.
5.2 Disinformation
In this section we discuss how fake news has been defined and interpreted by researchers who are looking
for algorithmic based deception detection solution for the fake news epidemic. As discussed in (section 2.3)
fake news, with its ambiguity and misuse, is still the favorite term among scholars to represent the spread
of false information.
While the debate is proceeding on, several deception detection algorithms are using fake news. Out of the
16 papers 13 of them used ‘fake news’ to discuss information disorder. There are few reasons that make
fake news the favorite choice. Fake news is attractive (Difranzo and Gloria, 2018), sensational (Karadzhov
et al., 2017), buzzword (Tandoc et al., 2018). In our coding and analysis phases we found that fake news is
possessing the definitions of disinformation. The words and sentences used to define fake news are identical
with the definitions of disinformation. But we cannot conclude that fake news and disinformation are the
same. The definitions given by Fallis (2009, 2015) and others didn’t mention fake news as synonym for
disinformation. Newman et al. (2018) argued that there is no agreed upon distinction between fake news,
ideologically slanted news, disinformation, misinformation and propaganda. Wardle (2017) defines
disinformation as deliberate creation and sharing of information known to be false. This and other
definitions put disinformation close to propaganda.
Martin (1982) claims that disinformation was derived from the Russian word ‘Dezinformatsiya’, which
defines as the dissemination of false and provocative information. Bittman (1990) agrees with the adaptation
of the word by the former soviets but he disported on the origin of the word. He claims disinformation has
its roots from German. Both Bittman and Martin classified disinformation as a form of propaganda. Fallis
(2009), who published papers on disinformation defined it as forged documents, doctored photographs,
deceptive advertising, deliberately falsified map and government propaganda. His definition has similar
themes as Tandoc et al., (2018) ‘typology of fake news’. Disinformation can adequately represent than fake
news. Wardle and Derakhshan (2018) stated the intent of fabricating and disseminating disinformation is to
cause harm to a person, social group, organizations or country. The disinformation definitions we have
discussed stem from the argument of Publications.europa.eu. (2018). This publication of European
33
Commission strongly argues that the problem is disinformation not fake news. Recently, the British
parliamentary committee published a report about threats of democracy. The committee concluded that the
term ‘fake news’ should not be used in the future, citing its misuse and lack of agreed definitions. The
committee suggested that the British government should use misinformation and disinformation instead
(Waterson, 2018). After reviewing definitions of fake news on 34 scholarly articles, Tandoc et al. (, 2018)
concluded that it is an oxymoron to use ‘fake news’ as a term. They argue that ‘news’ represents accurate
account of real events and ‘fake’ is for forgery, counterfeit and inauthentic.
Hoax has similarities with disinformation since all the definition features are the same, except hoax uses
humor to deceive. Hoax is defined as deceptive, fabricated and fictious (Concise Oxford Dictionary;
Wikipedia.com). this is similar to the features of disinformation.
We decided to pick disinformation instead of fake news for the following reasons;
1) we believe fake news lost its academic understanding,

2) fake news is inadequate and ambiguous, and
3) all the attributes listed as definitions of fake news are representative of disinformation.
Based on the thematic analysis we conducted (on section 4.2.2) we identified the following five terms as
the constructs of disinformation;
1) Being deliberate fabrication,
2) Being intentionally false,
3) Being digital deception,
4) Being a falsification, and
5) Being misleading.
5.2.1 Deliberate fabrication
Tandoc et al. (2018) defined fabrication as “articles which have no factual basis but are published in the
style of news articles”. Papers 2, 3, 4, 7, 9, 11, and 14 used deliberate fabrication and serious fabrication to
define fake news. Martin (1982), argued that disinformation is a form of propaganda based on unlawful act
or the intent to mislead and deceiving. Bittman (1990) and Martin (1982) stated disinformation as a critical
component of propaganda. Disinformation and propaganda have fabrication and the intent to deceive as a
common identifier.
5.2.2 Intentionally false
Intentionally false, which is a feature of disinformation, has been defined as inaccurate information that is
spread deliberately (Kragh and Åsberg, 2017). Svärd and Rumman (2017) termed it as propaganda. Papers
5 and 15 used the term to define fake news.
34
5.2.3 Digital deception
Social media which are often seen as disseminators of deceptive, distorted and false information, are blamed
as a breeding ground for ideological and hate propaganda. In their large-scale study of definitions of fake
news, Tandoc et al. (, 2018) interpreted deception as the degree to which the creator of fake news intends to
mislead. Levine (2014) defined deception as intentionally, knowingly and/or purposeful misleading another
person. Deception used on Paper 7 and 16. Paper 7 mentioned digital deception twice. In the first one it used
as part of fake news definition. In the latter it was defined as digital deception. Only paper 16 used the term
deception.
5.2.4 Falsification
Falsification is a term used in the definitions of disinformation and propaganda (Martin, 1982). But the
persuasive message it carries is more related to propaganda. Paper 2 used falsification twice to define fake
news and hoax. Paper 5 and 15 used intentional false information. Paper 7 used falsified to define fake news.
5.2.5 Misleading
Misleading is the only sub-theme that disinformation and misinformation have in common. Misleading can
be regarded as the end result of all the deception activities. Misleading is targeting the audience (Reilly,
2018) to gain profit (Publications.europa.eu., 2018). Misleading was mentioned in paper 6, 7, and 13,
together with fake news definitions. Paper 13 defined fake news a misleading for financial and political
gain. This definition equated misleading with deception.
5.3 Misinformation
The initial codes pulled out from the data (section 4.2.2) show the problem with defining fake news. A fake
news definition presented in paper 7 used deliberate misinformation with other disinformation features like
fabricated, falsified and digital deception. Researchers stressed that misinformation is a honest mistake or
negligence (Haiden et al., 2018; Bittman, 1990). It is unclear how honest mistake can be defined in
conjunction with a deliberate act. Paper 8, after having defined misinformation as unintentional spread of
false information, listed fake news and hoax as misinformation. Paper 14 and 16 defined misinformation as
honest mistake, unintentionally false and wrong information.
Misinformation can be regarded as unintentional dissemination or sharing of fake news (Wardle, 2017;
Kragh and Åsberg, 2018). Based on the analysis we tend to define it is an honest mistake without the intent
to deceive (Fallis, 2017).
We identified misinformation as a theme with the following four sub-themes.
1) Honest mistake 3) Incorrect facts

2) Wrong information 4) Misleading
35
5.3.1 Honest mistake
Honest mistake can be defined as an error or unintentional mistake. It can be termed as genuine mistake,
omission, prejudice or sheer ignorance (Bittman, 1990). Paper 14 and 16 used ‘mistake’ to define
misinformation.
5.3.2 Wrong information

Bittman (1990) explained that false information can be used as wrong information. On grounds of deliberate
action false news can be deceiving but without the intent it is wrong information (misinformation), (Haiden
et al., 2018). Unintentional false information can be termed misinformation (Wardle, 2017; Kragh and
Åsberg, 2017). Paper 16 defined misinformation as mistakenly providing wrong information. The mixed
use of false information as intentionally false and true false seems a bit confusing, but it can be easily
identified.
5.3.3 Incorrect facts

Incorrect facts have a close similarity to honest mistake as long as it does not have the intent to deceive.
Paper 14 described incorrect facts as a manifestation of misinformation.
5.3.4 Misleading
Misleading is the only sub-theme that disinformation and misinformation have in common. Mislead is
defined as “to lead in a wrong direction or into a mistaken action or belief often by deliberate treachery”
(Meriamwebester.com). Misleading is the output of intentional actions of disinformation and it is the
unwanted or undesired effect of misinformation. Fallis (2015) stated that misinformation does not have the
intent to deceive but it is misleading. A European Union publication (2018) on disinformation defined
misinformation as misleading and inaccurate information. in their definitions of misinformation paper 8 and
14 failed to include misleading as an effect of misinformation.
5.4 Definitions from the papers
This section presents the definitions we collected from the 16 scholarly articles. The analysis shows that
most papers use definitions of fake news that are inadequate and unsatisfactory for research purposes.
Paper 1: Conroy et al. (2015)
Definition(s):
Fake news is Intentionally deceptive news report.
Theme Sub-theme found in definition

Disinformation Misleading
Misinformation
Table 6 Paper 1: Definition Sub-themes
36
Paper 2: Rubin et al. (2015)
Definition(s):
Three types of fake news;
1) Serious fabrication, Fraudulent reporting, falsification, or exaggeration uncovered in mainstream
media or participant media, yellow press or tabloids.
2) Hoaxing is another type of deliberate fabrication or falsification in the mainstream or social media.
3) Humorous fakes “in a format typical of mainstream journalism but rely heavily on irony and deadpan
humor to emulate a genuine news source, mimicking credible news sources and stories, and often
achieving wide distribution (news satire, parody, game shows).

Disinformation Fabrication
Falsification
Misinformation
Paper 3: Tacchini et al. (2017)
Definition(s):
Hoax is Intentionally crafted fake information.

Misinformation
Paper 4: Ruchansky et al. (2017)
Definition(s):
Fake news is News content that is fabricated.

Disinformation fabrication
Misinformation
Paper 5: Shu et al. (2017)
Definition(s):
Fake news is low quality news with intentionally false information.

Disinformation Intentionally false
Misinformation
37
Paper 6: Ahmed et al. (2018)
Definition(s):
Three fake news groups;
False news- which is news that is completely fake and is made up by the writers of the articles.
Fake satire news- which is fake news whose main purpose is to provide humor to the readers.
Poorly written news article- which have a certain degree of real news but are not entirely accurate.
Misinformation
Paper 7: Rubin and Chen (2015)
Definition(s):
Fake news-Fake, fabricated, falsified, disingenuous, or misleading news reports constitute instances of
digital deception or deliberate misinformation.
Digital deception-intentional control of information in a technologically mediated environment to create a
false belief or false conclusion.
Falsification
Misleading
Digital deception
Misinformation
Paper 8: Shao et al. (2016)
Definition(s):
Misinformation is unintentional spread of false or inaccurate information. misinformation are rumors,

hoaxes, fake news, and conspiracy theories stated as examples of misinformation.
Disinformation
Incorrect facts
Wrong information
Paper 9: Perez-Rosas et al. (2017)
Definition(s):
Fake news- serious fabrications i.e. news items about false and nonexistent events or information such as
celebrity gossip.
38
Misinformation
Paper 10: Tschitschek et al. (2018)
Definition(s):
Fake news also known as hoax, rumors etc.

Disinformation
Misinformation
Paper 11: Thorne et al. (2017)
Definition(s):
Fake news is highly partisan fabricated materials on social media.

Misinformation
Paper 12: Pfohl et al. (2017)
Definition(s):
Fake news is a made-up story with an intention to deceive.

Misleading
Misinformation
Paper 13: Karadzhov et al. (2018)
Definition(s):
Fake news is written and published with the intent to mislead to gain financially or politically, often targeting
specific user groups.

Misinformation
39
Paper 14: Kumar et al. (2016)
Definition(s):
Disinformation, Hoax, and misinformation:

Misinformation- is conveyed in the honest but mistaken belief that the relayed incorrect facts are true.
Misinformation- information that is unintentionally false.
Disinformation- false facts that are conceived to deliberately deceive or betray an audience.
Hoax is a kind of disinformation.
Hoax- A deliberately fabricated falsehood made to masquerade as truth.

Falsification
Incorrect facts
Paper 15: Shu et al. (2018)
Definition(s):
Fake news is news with intentionally false information.

Disinformation Intentionally false
Misinformation
Paper 16: Santos and Li (2010)
Definition(s):
Deception and misinformation
Deception as information designed to manipulate the behavior of others by inducing them to accept a false
or distorted presentation of their environment—physical, social, or political.
Deception as a deliberate act perpetrated by a sender to engender in a receiver’s beliefs contrary to what
the sender believes is true to put the receiver at a disadvantage.
misinformation is defined as mistakenly providing the wrong information.

Misinformation Wrong information
40
6 Discussion
Fake news detection has become much discussed topic. Fact checking, crowdsourcing and algorithmic
detection methods find themselves found in the midst of those debates. We found few research articles that
propose algorithmic deception detection as a remedy. Out of those few, there is an even smaller number
with definitions of fake news. Those found with definitions are incomplete and do not present a reasoned
understanding. Conceptually, the definitions are not well thought through. In any problem-solving
mechanism problem definition is the first task that needs to be performed methodically. The study shows
that fake news is incorrectly defined and understood.
Combating online fake news with algorithmic deception detection is a relatively young phenomenon. Ten
Out of the 16 papers were produced in 2017 and 2018. One of the reasons for the unusually insufficient
definition of fake news might be the datasets on which the algorithms were trained. Out of the 16 papers
only 1 claimed to have its own dataset.
Out of 16 only 7 papers satisfactorily defined the problem they are trying to tackle. Nine of them defined
fake news in just one line using very few words. We found one paper that used only two words to define
fake news.
For those who tried to define fake news, disinformation, hoax or misinformation the lack of agreed upon
definitions might be a hinderance. The ambiguity and the misuse of fake news can be another reason. Tandoc
et al. (2018) explained why we need acceptable fake news definition: “A clear definition of fake news, one
that matches its imperial manifestation, can help in testing and building theories in news production”.
Researchers from linguistic, journalism, computer science and other fields of study need to work together
to come up with a definition that can be accepted as an academic definition.
Computer science researchers keep using the term ‘fake news’ while others are abandoning it. Governments,
institutions and European union publications are refraining from using fake news citing its ambiguity and
misuse.
Our results have shown that disinformation can replace fake news. Most of the academic meanings given to
fake news belongs to disinformation. To clear the confusion created by the many faces of fake news,
disinformation is the most suitable replacement.
41
7 Conclusion
Detecting Fake news in the era of social media without technological intervention is unthinkable. Fact
checking sites, fake website identifiers, applications, plugins and deception detection algorithms are helping
to combat fake news. They are being used to shield social liberties exemplified by elections from foreign
interference and intrusion. The internet and, mainly the social media are the new battleground for an
information war. A country, a business or a person can lose fortunes by a single bit of false information.
Candidates for political office are getting elected with the help of fabricated and falsified information.
Studying how fake news, disinformation, hoax and misinformation are defined for the purpose of
algorithmic deception detection has challenges. In our research we were able to identify only 16 recent
publications trying to define fake news. Only 7 out of 16 papers satisfactorily defined the problem they are
trying to tackle. The result shows that algorithmic based research papers have a great difficulty in defining
fake news, disinformation, hoax and misinformation, and even when they produce a definition it is
unsatisfactory and the two hypotheses (section 4.1) proved to be true.
Algorithmic solutions have a huge potential in combating fake news and other information disorder. In
regard to this research the deception detection algorithmic solutions fail short to deliver the information
needed for the analysis. If more than half of the papers examine defined fake news inadequately, one should
wonder about how successful their algorithms are in identifying and detecting false information.
7.1 Future research

False news fabrication and falsification are on the upswing and becoming increasingly complex. National
states are involved in fabricating and disseminating false information by utilizing their human and
technological power. This interference showed its potential on the 2016 US election and during Brexit
referendum.
To minimize the attacks and protect civil liberties a new approach is needed. Most of the algorithmic
deception detection solutions are engaged on detecting the semantic meaning of words. That is not enough.
Combating false information requires algorithms that can identify the context and the hidden meanings in
words, and therefore a search for pragmatic meaning.
Understanding and utilizing the pragmatics of meaning demand a novel approach and techniques on top of
the usual training of algorithms on existing datasets. Pragmatics analysis can create a lot of room for future
research in the field.
Several deception detection algorithms are being developed to combat fake news. To successfully combat
fake news, the phenomena needs to be well defined, categorized and understood.
This work can be used as a stepping stone by future researchers who are interested to develop better
performing fake news deception detection algorithms because of a better definition and understanding of
the problem.
42
Bibliography
1. Ahmed, H., Traore, I. and Saad, S., 2018. Detecting opinion spams and fake news using text
classification. Security and Privacy, 1(1), p.e9.
2. Allcott, H. and Gentzkow, M., 2017. Social media and fake news in the 2016 election. Journal of
Economic Perspectives, 31(2), pp.211-36.
3. Bakir, V. and McStay, A., 2018. Fake news and the economy of emotions: Problems, causes,
solutions. Digital Journalism, 6(2), pp.154-175.
4. Berghel, H., 2017. Lies, damn lies, and fake news. Computer, (2), pp.80-85.
5. Bittman, L., 1990. The use of disinformation by democracies. International Journal of Intelligence
and Counter Intelligence, 4(2), pp.243-261.
6. Braun, V. and Clarke, V., 2006. Using thematic analysis in psychology. Qualitative research in
psychology, 3(2), pp.77-101.
7. Braun, V., Clarke, V. and Terry, G., 2014. Thematic analysis. Qual Res Clin Health Psychol, 24,
pp.95-114.
8. Brownlee, J. (2018). How to Define Your Machine Learning Problem. [online] Machine Learning
Mastery. Available at: https://machinelearningmastery.com/how-to-define-your-machine-
learning-problem/ [Accessed 29 Jul. 2018].
9. Burgoon, J.K. and Buller, D.B., 1994. Interpersonal deception: III. Effects of deceit on perceived
communication and nonverbal behavior dynamics. Journal of Nonverbal Behavior, 18(2), pp.155-
184.
10. Chomsky, N., 1957. Syntactic structures. The Hague: Mouton. 1965. Aspects of the theory of
syntax. Cambridge, Mass.: MIT Press. (1981) Lectures on Government and Binding, Dordrecht:
Foris. (1982) Some Concepts and Consequences of the Theory of Government and Binding. LI
Monographs, 6, pp.1-52.
11. Collinsdictionary.com. (2018). Collins - The Collins Word of the Year 2017 is.... [online] Available
at: https://www.collinsdictionary.com/woty [Accessed Feb15. 2018].
12. Confessore, N., Dance, G.J., Harris, R. and Hansen, M., 2018. The follower factory. The New York
Times. January, 27.
13. Conroy, N.J., Rubin, V.L. and Chen, Y., 2015, November. Automatic deception detection: Methods
for finding fake news. In Proceedings of the 78th ASIS&T Annual Meeting: Information Science
with Impact: Research in and for the Community (p. 82). American Society for Information Science.
14. Cormen, T.H., Leiserson, C.E., Rivest, R.L. and Stein, C., 2009. Introduction to algorithms. MIT
press.
43
15. Darnton, R. (2018). The True History of Fake News. [online] The New York Review of Books.
Available at: https://www.nybooks.com/daily/2017/02/13/the-true-history-of-fake-news/
[Accessed 30 Jul. 2018].
16. DiFranzo, D. and Gloria-Garcia, K., 2017. Filter bubbles and fake news. XRDS: Crossroads, The
ACM Magazine for Students, 23(3), pp.32-35.
17. Fallis, D., 2009. A conceptual analysis of disinformation.
18. Fallis, D., 2015. What is disinformation?. library trends, 63(3), pp.401-426.
19. Ferrara, E., Varol, O., Davis, C., Menczer, F. and Flammini, A., 2016. The rise of social
bots. Communications of the ACM, 59(7), pp.96-104.
20. Flood, A. (2016), ‘post-truth’ named word of the year by oxford dictionaries. Retrieved from
https://www.theguardian.com/2016/nov/15
21. Francis, p. (2018). Pope Francis releases 2018 World Communications Day message - Vatican
News. [online] Vaticannews.va. Available at: https://www.vaticannews.va/en/pope/news/2018-
01/pope-world-communications-day-message-2018-truth-journalism-fake.html [Accessed 29 Jul.
2018].
22. Gilda, S., 2017, December. Evaluating machine learning algorithms for fake news detection. In
Research and Development (SCOReD), 2017 IEEE 15th Student Conference on (pp. 110-115).
IEEE.
23. Grice, A. (2018). Fake news handed Brexiteers the referendum – and now they have no idea what
they're doing. [online] The Independent. Available at: http://www.independent.co.uk /voices/
michael-gove-boris-johnson-brexit-eurosceptic-press-theresa-may-a7533806.html [Accessed
Feb15. 2018].
24. Gu, L., Kropotov, V. and Yarochkin, F., 2017. The fake news machine, how propagandists abuse
the internet and manipulate the public. pdf] Trend Micro, 81, pp.1073547711-1497355570.
25. Haiden, L., McManus, C., Michaud, C., Duffy, E., Ranautta-Sambhi, K. and Ilbury, M., 2018. A
ROADMAP.
26. Haiden, L. and Althuis, J., 2018. The Definitional Challenges of Fake News.
27. Hofmann, T.R., 2015. Realms of meaning: an introduction to semantics. Routledge.
28. Kaplan, B. and Maxwell, J.A. “Qualitative Research Methods for Evaluating Computer Information
Systems,” in Evaluating Health Care Information Systems: Methods and Applications, J.G.
Anderson, C.E. Aydin and S.J. Jay (eds.), Sage, Thousand Oaks, CA, 1994, pp. 45-68.
29. Karadzhov, G., Gencheva, P., Nakov, P. and Koychev, I., 2018. We Built a Fake News & Click-
bait Filter: What Happened Next Will Blow Your Mind!. arXiv preprint arXiv:1803.03786.
44
30. Kragh, M. and Åsberg, S., 2017. Russia’s strategy for influence through public diplomacy and
active measures: the Swedish case. Journal of Strategic Studies, 40(6), pp.773-816.
31. Kumar, S., West, R. and Leskovec, J., 2016, April. Disinformation on the web: Impact,
characteristics, and detection of wikipedia hoaxes. In Proceedings of the 25th international
conference on World Wide Web (pp. 591-602). International World Wide Web Conferences
Steering Committee.
32. Larson, R.K. and Segal, G., 1995. Knowledge of meaning: An introduction to semantic theory (pp.
310-11). Cambridge, MA: Mit Press.
33. Lazer, D., Baum, M., Grinberg, N., Friedland, L., Joseph, K., Hobbs, W. and Mattsson, C., 2017.
Combating fake news: An agenda for research and action. Harvard Kennedy School, Shorenstein
Center on Media, Politics and Public Policy, 2.
34. Lazer, D.M., Baum, M.A., Benkler, Y., Berinsky, A.J., Greenhill, K.M., Menczer, F., Metzger,
M.J., Nyhan, B., Pennycook, G., Rothschild, D. and Schudson, M., 2018. The science of fake
news. Science, 359(6380), pp.1094-1096.
35. Levine, T.R., 2014. Truth-Default Theory (TDT) A theory of human deception and deception
detection. Journal of Language and Social Psychology, 33(4), pp.378-392.
36. Lewandowsky, S., Ecker, U.K. and Cook, J., 2017. Beyond misinformation: Understanding and
coping with the “post-truth” era. Journal of Applied Research in Memory and Cognition, 6(4),
pp.353-369.
37. Marsden, C., 2017. How law and computer science can work together to improve the information
society. Communications of the ACM, 61(1), pp.29-31.
38. Martin, L.J., 1982. Disinformation: An instrumentality in the propaganda arsenal. Political
Communication, 2(1), pp.47-64.
39. Merriam-webster.com. (2018). Definition of HOAX. [online] Available at: https://www.merriam-
webster.com/dictionary/hoax [Accessed 5 Aug. 2018].
40. Mustafaraj, E. and Metaxas, P.T., 2017, June. The fake news spreading plague: was it preventable?.
In Proceedings of the 2017 ACM on Web Science Conference (pp. 235-239). ACM.
41. Myers, M.D., 1997. Qualitative research in information systems. Management Information Systems
Quarterly, 21(2), pp.241-242.
42. Newman, N., Fletcher, R., Kalogeropoulos, A., Levy, D.A. and Nielsen, R.K., 2017. Reuters
Institute digital news report 2017.
43. Oxford Dictionaries | English. (2018a). hoax | Definition of hoax in English by Oxford Dictionaries.
[online] Available at: https://en.oxforddictionaries.com/definition/hoax [Accessed 6 Aug. 2018].
45
44. Oxford Dictionaries | English. (2018b). post-truth | Definition of post-truth in English by Oxford
Dictionaries. [online] Available at: https://en.oxforddictionaries.com/definition/post-truth
[Accessed 11 Aug. 2018].
45. Pérez-Rosas, V., Kleinberg, B., Lefevre, A. and Mihalcea, R., 2017. Automatic Detection of Fake
News. arXiv preprint arXiv:1708.07104.
46. Persily, N., 2017. The 2016 US Election: Can democracy survive the internet?. Journal of
democracy, 28(2), pp.63-76.
47. Pfohl, S., Triebe, O. and Legros, F., 2017. Stance Detection for the Fake News Challenge with
Attention and Conditional Encoding.
48. Publications.europa.eu. (2018). A multi-dimensional approach to disinformation: report of the
independent High-level Group on fake news and online disinformation.. [online] Available at:
https://publications.europa.eu/en/publication-detail/-/publication/6ef4df8b-4cea-11e8-be1d-
01aa75ed71a1/language-en [Accessed 30 Jul. 2018].
49. Reilly, I., 2018. F for Fake: Propaganda! Hoaxing! Hacking! Partisanship! and Activism! in the
Fake News Ecology. The Journal of American Culture.
50. Reuters.com (2018). Factbox: Action by Asian governments to crack down on 'fake news'. [online]
U.S. Available at: https://www.reuters.com/article/us-asia-fakenews-factbox/factbox-action-by-
asian-governments-to-crack-down-on-fake-news-idUSKCN1HA0QF [Accessed 1 Aug. 2018].
51. Research-Methodology. (2018). Exploratory Research - Research-Methodology. [online] Available
at:https://research-methodology.net/research-methodology/research-design/exploratory
research/#_ftn1 [Accessed 10 Aug. 2018].
52. Rochlin, N., 2017. Fake news: belief in post-truth. Library Hi Tech, 35(3), pp.386-392.
53. Rose, J., 2017. Brexit, trump, and post-truth politics.
54. Rubin, V.L., Chen, Y. and Conroy, N.J., 2015. Deception detection for news: three types of
fakes. Proceedings of the Association for Information Science and Technology, 52(1), pp.1-4.
55. Rubin, V.L., Conroy, N.J. and Chen, Y., 2015, January. Towards news verification: Deception
detection methods for news discourse. In Proceedings of the Hawaii International Conference on
System Sciences (HICSS48) Symposium on Rapid Screening Technologies, Deception Detection
and Credibility Assessment Symposium, January (pp. 5-8).
56. Ruchansky, N., Seo, S. and Liu, Y., 2017, November. Csi: A hybrid deep model for fake news
detection. In Proceedings of the 2017 ACM on Conference on Information and Knowledge
Management (pp. 797-806). ACM.
57. Santos, E. and Li, D., 2010. On deception detection in multiagent systems. IEEE Transactions on
Systems, Man, and Cybernetics-Part A: Systems and Humans, 40(2), pp.224-235.
46
58. Shao, C., Ciampaglia, G.L., Flammini, A. and Menczer, F., 2016, April. Hoaxy: A platform for
tracking online misinformation. In Proceedings of the 25th international conference companion on
world wide web (pp. 745-750). International World Wide Web Conferences Steering Committee.
59. Shu, K., Sliva, A., Wang, S., Tang, J. and Liu, H., 2017. Fake news detection on social media: A
data mining perspective. ACM SIGKDD Explorations Newsletter, 19(1), pp.22-36.
60. Shu, K., Bernard, H.R. and Liu, H., 2018. Studying Fake News via Network Analysis: Detection
and Mitigation. arXiv preprint arXiv:1804.10233.
61. Sille Obelitz Søe, (2018) "Algorithmic detection of misinformation and disinformation: Gricean
perspectives", Journal of Documentation, Vol. 74 Issue: 2, pp.309-332, https://doi.org/10.1108/JD-
05-2017-0075
62. Smith, A. (2018). Franken-algorithms: the deadly consequences of unpredictable code. [online] the
Guardian. Available at: https://www.theguardian.com/technology/2018/aug/29/coding-algorithms-
frankenalgos-program-danger [Accessed 6 Sep. 2018].
63. Stelter, B. (2018). Trump averages a 'fake' insult every day. Really. We counted. [online]
CNNMoney. Available at: http://money.cnn.com/2018/01/17/media/president-trump-fake-news-
count/index.html [Accessed Feb15. 2018].
64. Suiter, J., 2016. Post-truth politics. Political Insight, 7(3), pp.25-27.
65. Svärd, M. and Rumman, P., 2017. COMBATING DISINFORMATION: Detecting fake news with
linguistic models and classification algorithms.
66. Sydell, L., 2016. We tracked down a fake-news creator in the suburbs. here’s what we
learned. National Public Radio, 23.
67. Tacchini, E., Ballarin, G., Della Vedova, M.L., Moret, S. and de Alfaro, L., 2017. Some like it hoax:
Automated fake news detection in social networks. arXiv preprint arXiv:1704.07506.
68. Tandoc Jr, E.C., Lim, Z.W. and Ling, R., 2018. Defining “fake news” A typology of scholarly
definitions. Digital Journalism, 6(2), pp.137-153.
69. Thelocal.se. (2018). How Sweden’s getting ready for the election year information war. [online]
Available at: https://www.thelocal.se/20171107/how-swedens-getting-ready-for-the-election-year-
information-war [Accessed Feb15. 2018].
70. Thomas, J.A., 2014. Meaning in interaction: An introduction to pragmatics. Routledge.
71. Tschiatschek, S., Singla, A., Gomez Rodriguez, M., Merchant, A. and Krause, A., 2018, April. Fake
News Detection in Social Networks via Crowd Signals. In Companion of the Web Conference 2018
on The Web Conference 2018 (pp. 517-524). International World Wide Web Conferences Steering
Committee.
47
72. Vosoughi, S., Roy, D. and Aral, S., 2018. The spread of true and false news online. Science,
359(6380), pp.1146-1151.
73. Wardle, C., 2017. Fake news. It’s complicated. First Draft News.
74. Wardle, C. and Derakhshan, H. (2018). Information Disorder. [online] Rm.coe.int. Available at:
https://rm.coe.int/information-disorder-toward-an-interdisciplinary-framework-for-
researc/168076277c [Accessed 2 Aug. 2018].
75. Waterson, J. (2018). Democracy at risk due to fake news and data misuse, MPs conclude. [online]
the Guardian. Available at: https://www.theguardian.com/technology/2018/jul/27/fake-news-
inquiry-data-misuse-deomcracy-at-risk-mps
conclude?utm_source=esp&utm_medium=Email&utm_campaign=GU+Today+main+NEW+H+c
ategories&utm_term=282232&subid=15992303&CMP=EMCNEWEML6619I2 [Accessed 29 Jul.
2018].
76. Zhou, L., Burgoon, J.K., Nunamaker, J.F. and Twitchell, D., 2004. Automating linguistics-based
cues for detecting deception in text-based asynchronous computer-mediated
communications. Group decision and negotiation, 13(1), pp.81-106.
48

Defining Fake News For Algorithmic Deception Detection

Uploaded by

Copyright:

Available Formats

Defining Fake News For Algorithmic Deception Detection

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Defining Fake News For Algorithmic Deception Detection

Uploaded by

Copyright:

Available Formats

UPPSALA

Defining fake news for algorithmic deception detection

Yohannes Kassahun Asresu

Subject: Information Systems

Department of Informatics and Media

Keywords: Fake news, Disinformation, Deception detection, Algorithm, Misinformation, Information

2 What is Fake News? 9

An overview of fake news

1.2 Research problem

1. How well do algorithms define fake news?

1.4 Aim of the study

2.1 Fake News- to use it or not

2.2 Fake news, post-truth and Gaudy Facts

2.3.1 Defining fake news

2.3.2 Typology of Fake News

Fig 1. Typology of misinformation and disinformation (Wardle, 2017)

(1) news that is ‘invented’ to make money or discredit others;

- “False, inaccurate, or misleading information designed, presented and promoted to intentionally

- “Disinformation is intentionally false or inaccurate information that is spread deliberately” (Kragh

- “Disinformation is nothing new, of course. Forged documents, doctored photographs, deceptive

- Dis-information- “Information that is false and deliberately created to harm a person,

- “Misinformation is unintentional dissemination of false information” (Kragh and Åsberg, 2017).

- “Misinformation refers to the inadvertent sharing of false information” (Wardle, 2017).

- “Misinformation is inaccurate information that is the result of an honest mistake or of negligence”

- “Misleading or inaccurate information shared by people who do not recognize it as such”

- “Propaganda is something that is conducted by one's opponents, and it must be unsuccessful to be

- Hoaxing as “a humorous or mischievous deception, usually taking the form of a fabrication of

- “A hoax is a falsehood deliberately fabricated to masquerade as the truth” (Wikipedia.com).

(2) Language understanding - contains context-dependent features of utterance meaning. relevant

3.3 Related Work

A) Fake news is badly or inadequately defined.

4.1.1 Mapping Definitions

Structure of the mapping

4.2 Coding and analysis of definitions

What is considered as a theme?

4.2.2 Doing the analysis

Phase 1: knowing your data, Phase 4: Reviewing themes,

Phase 2: Generate initial codes

Article Data items Initial codes

Phase 3: Searching for themes

Phase 4: Reviewing themes

Disinformation Deliberate fabrication

Falsification Deliberate Honest Wrong

Intentionally Digital Incorrect

Fig 2: Final Thematic Map

5.1 Identified Themes

1) we believe fake news lost its academic understanding,

5.2.1 Deliberate fabrication

5.2.2 Intentionally false

We identified misinformation as a theme with the following four sub-themes.

1) Honest mistake 3) Incorrect facts

5.3.2 Wrong information

5.3.3 Incorrect facts

5.4 Definitions from the papers

Paper 1: Conroy et al. (2015)

Fake news is Intentionally deceptive news report.

Theme Sub-theme found in definition

Theme Sub-theme found in definition

Paper 3: Tacchini et al. (2017)