Conference PaperPDF Available

CrowdsouRS: A Crowdsourced Reputation System for Identifying Deceptive Online Contents



In recent years, accelerated web-based technologies have revolutionized content generation and broadcast mecha- nisms through the Internet. Social media, blogs, e-newspaper, auction sites facilitate the creation and exchange of user-generated contents, which rarely go through any fact-finding mechanism or rigorous editorial process. This has fuelled the creation and publication of fake news in the web. The proliferation of social networks has been exploited to accelerate the distribution and propagation of such fake news at an unprecedented level, creating a major concern for the web. There have been several efforts undertaken to rectify this problem, unfortunately, none seems to be effective to root out this concerning issue. In this paper, we present CrowdsouRS, a Crowd-sourced Reputation System, implemented as a browser extension, for the web that leverages the wisdom of the crowd to identify and tag deceptive online contents. It aggregates reputation scores for a web page from multiple users, which is then visualized in order to help other users to determine if the contents of the web page are deceptive. We have evaluated the usability and effectiveness of CrowdsouRS with a number of users and our evaluations suggest that users find the tool useful in serving its purpose.
To be published in the 20th International Conference of Computer and Information Technology, 2017, (ICCIT’17)
CrowdsouRS: A Crowdsourced Reputation System
for Identifying Deceptive Online Contents
Masiur Rahman Siddiki
Department of CSE, SUST
Md. Abu Talha
Department of CSE, SUST
Farida Chowdhury
Department of CSE, SUST
Md Sadek Ferdous
Imperial College London, London
Abstract—In recent years, accelerated web-based technologies
have revolutionized content generation and broadcast mecha-
nisms through the Internet. Social media, blogs, e-newspaper,
auction sites facilitate the creation and exchange of user-generated
contents, which rarely go through any fact-finding mechanism
or rigorous editorial process. This has fuelled the creation and
publication of fake news in the web. The proliferation of social
networks has been exploited to accelerate the distribution and
propagation of such fake news at an unprecedented level, creating
a major concern for the web. There have been several efforts
undertaken to rectify this problem, unfortunately, none seems
to be effective to root out this concerning issue. In this paper,
we present CrowdsouRS, a Crowd-sourced Reputation System,
implemented as a browser extension, for the web that leverages
the wisdom of the crowd to identify and tag deceptive online
contents. It aggregates reputation scores for a web page from
multiple users, which is then visualized in order to help other
users to determine if the contents of the web page are deceptive.
We have evaluated the usability and effectiveness of CrowdsouRS
with a number of users and our evaluations suggest that users
find the tool useful in serving its purpose.
KeywordsTrust, Reputation, Browser extension, Fake news,
Crowd sourcing
The act of employing deceptions to gain illicit advantages
is as old as the human society. A deceptive act can be
characterized from inaccurate statements to misleading claims
which hide or skip relevant information, leading the receiver
to assume false conclusions. With the advent of electronic-
mediated communication systems, it is much easier to carry
out a deceptive act with potentially greater adversarial effects.
One prominent example within this category is Fake News
published and disseminated via web.
In recent years, accelerated web-based technologies have
revolutionized content generation and broadcast mechanisms
through the Internet. Social media, blogs and e-newspaper
facilitate the creation and exchange of user-generated contents,
which rarely go through any fact-finding mechanism or rig-
orous editorial process. The proliferation of these mediums
has fueled the creation and propagation of fake news at an
unprecedented level, creating a major concern for the web.
A recent incident showcasing the detrimental effect of fake
news was when Facebook encountered criticisms that it might
have helped President Donald Trump get elected via a massive
propagation of fake news on the platform. In light of this
event, Facebook has begun tagging fake news using a fact-
checking service called [1], [2]. However,
the method is not effective yet [3]. In a similar vein, Google
is taking steps to prevent fake news in its search results to
tackle this problem [4], [5]. They are using fact-checking labels
which have minimal impact, sometimes results in the opposite
In Bangladesh, people are increasingly being faced with
deceptive contents in the form of fake news, rumours and
viral Facebook posts. One example is that a famous newspaper
of Bangladesh named Kaler Kantho published a fake news
with hardly any basis, where it was reported that the editor
of another newspaper ”Prothom-Alo”, Matiur Rahman was
involved in 2004 grenade assassination attempt against the
Bangladeshi Prime minister [7]. Moreover, we have witnessed
several racially and religiously aggravated attacks based on
falsified Facebook posts and sharing [8]. This has motivated
us to carry out research to tackle this ever-increasing concern.
However, unlike the approaches adopted by different online
service providers such as Facebook and Google, we have
adopted a considerably different approach. We have explored
how the wisdom of crowd can be leveraged to identify decep-
tive contents in the web. Several well-established researches
have demonstrated the effectiveness of crowd wisdom in mul-
titude of domains [9], [10]. In this paper, we present Crowd-
souRS, a Crowd-sourced Reputation System, which leverages
the wisdom of crowd for building a reputation system whose
primary objective is to help a user to identify deceptive online
Contributions:In this paper, our contributions are as follows:
We present the CrowdsouRS, a browser-extension
based Trust and Reputation system for the web.
We present its architecture and implementation details
to illustrate our design choices that make it easily
We present a thorough user-study to test the usability
and applicability of our system.
Structure:We provide a brief background on trust and rep-
utation systems in Section II. In Section III, we present the
CrowdsouRS system along with its architecture, development
methodology and implementation details. Section IV presents
the user study. We evaluate the result and its implications in
Section V. We present a few related works in Section VI and
then finally, conclude in Section VII with a hint of future work.
In this section, we present a brief overview of the related
concepts of trust, reputation and their management.
2017 IEEE. Personal use of this material is permitted. Permission from IEEE must be obtained for all other uses, in any current or
future media, including reprinting/republishing this material for advertising or promotional purposes, creating new collective works, for resale or
redistribution to servers or lists, or reuse of any copyrighted component of this work in other works.
The concept of trust is a widely studied topic in different
domains which has been defined in numerous ways. In this
paper, we assume the following definition taken from [11]:
Trust is the extent to which one party is willing to depend
on something or somebody in a given situation with a feeling
of relative security, even though negative consequences are
The definition outlines a directional relationship between
two parties, a Trustor and a Trustee, where a trustor trusts a
trustee for something in a situation.
The reputation concept is closely related to the concept of
trust. Here, we assume the following definition of reputation
[12] where it is portrayed as a quality of an entity: “The opin-
ion that people in general have about someone or something,
or how much respect or admiration someone or something
receives, based on past behaviour or character.
The reputation concept is a widely deployed mechanism in
different web services such as Amazon, EBay, Netflix and so
on. In such systems, a reputation is often represented using a
score and is used as a collective measure of reliability/quality
for a product or content within these web services. To calculate
a reputation score, different mechanisms are available [13].
Among these mechanisms, we restrict our focus onto the
Multinomial Bayesian reputation system [14], [15], [16], [17]
as CrowdsouRS is based on this mechanism.
The reputation score in a Bayesian reputation system is cal-
culated by combining the previous score(s) with the new score
using either binomial Beta or multinomial Dirichlet probability
density functions (PDF) [18], [13]. It is of two types: Binomial
and Multinomial. A binomial system restricts a reputation
score to have only two values: positive or negative, whereas,
a reputation score in a multinomial system can be expressed
using a range of levels. In addition, a multinomial system can
distinguish between polarized and average ratings. In short,
a multinomial reputation system offers greater flexibility and
that is why it is often preferred over a binomial system.
Let us assume that a multinomial Bayesian reputation
system allows a reputation score to be expressed using k
different values, represented as a set of kdisjoint elements:
L={L1, L2, ..., Lk}. If we consider each rating provided by
a user as a vote on the element of L, a Dirichlet PDF over
the k-component random variable p(Li)can be defined with
sample space [0,1]k, where i= 1,2, ..., k.
The individual rating of each user for a particular target
(e.g. a service or a content) in the system is then aggregated
and the multinomial reputation score of the target, denoted as
the vector
S, is calculated using the following formula [15]:
Sy:sy(Li) = Ry(Li)+Ca(Li)
j=1 Ry(Lj);|i= 1...k(1)
Rrepresents the aggregate ratings of the target.
yrepresents a particular target/agent.
arepresents the base rate vector over L.
Crepresents a priori weight.
The reputation score
S, in essence, represents a probability
measurement of the target indicating how it might behave in
future when a newer reputation score is provided. However,
a more preferred approach to represent a reputation score is
to use the single-valued point estimate reputation for each
target. Such a score (denoted by σ) can be estimated using
the following formula [15].
i=1 v(i)S(Li)(2)
σrepresents the normalized weighted point estimate
v(i) = i1
k1represents a point value for each reputa-
tion score level.
In this section we present our system, called CrowdsouRS
- an acronym for Crowd-sourced Reputation System. In what
follows, we present its motivation (Section III-A), describe
its architecture & protocol flow (Section III-B) and discuss
its development and implementation methodologies (Section
A. Motivation
CrowdsouRS is empowered by the collective wisdom
of crowd, gathered by a crowd-sourcing approach. Crowd-
sourcing is a popular online activity used for providing product
ratings in online marketplaces such as Amazon, EBay and
Netflix. Backed by the crowd-sourcing approach, CrowdsouRS
employs the following assumption. People within any commu-
nity are the best judge to identify any deceptive online content
such as fake news. A deceptive content might deceive a small
group of people within a community. However, the collective
wisdom of the whole community can identify such content as
The wisdom of crowd in CrowdsouRS is represented using
a reputation score. The goal of CrowdsouRS is to enable every
user with a tool, while accessing a web page, that can be
utilized to provide a user rating regarding its contents. Each
provided rating will represent a single user’s perception of the
content. Every individual rating can then be aggregated using
a mechanism to transform it into a reputation score which will
signify the collective wisdom of a community.
B. Architecture
CrowdsouRS consists of two components (Figure 1): a
browser extension and a reputation server. The browser ex-
tension represents the front-end user interface by which a user
can interact with CrowdsouRS for providing a rating. Since the
primary focus of CrowdsouRS is to identify fake online news
accessed via a browser, the user interface has been designed
as a browser extension. Each web page in CrowdsouRS is
uniquely identified by its corresponding URL. The rating a
user provides for a web page is bound to the URL of the web
page. The browser extension is also responsible for displaying
the reputation score and reviews (if any) to the user.
The reputation server consists of a reputation calculation
engine and a database. The server provides the back-end
facility in which it interacts with the extension to collect the
individual rating from different users. It then aggregates all
ratings associated with a single URL to calculate the reputation
score for that web page. The calculated reputation is then
sent back to the extension for presenting via the extension.
In addition, the reputation score for the corresponding URL is
stored in the database.
Request Web Page URL of
Web Page
Web Server Reputation Server
Fig. 1: CrowdsouRS Architecture
C. Development & Implementation
CrowdsouRS extension has been developed for Chrome,
Opera or any other chromium engine based browser. It utilizes
HTML5, CSS & JavaScript framework JQuery for the user
interface. When the extension is clicked, a pop-up appears
(Figure 2). The pop-up is denoted as the home page of the ex-
tension which displays the reputation score of the current page
as returned by the reputation server, a review tab presenting
the review of other users and some other statistics regarding
the corresponding URL.
In addition, it shows a list of ratings that can be provided
by the user. The list contains five rating levels: Deceptive,
Untrustworthy, Average, Trustworthy and Very Trustworthy,
representing the elements of Las discussed in Section II.
Once a user selects and submits a particular rating, it is sent
to the reputation server through a POST HTTP request via
JQuery AJAX, where it is aggregated using the mechanism
discussed below. Furthermore, the pop-op also allows the user
to provide a textual review and feedback (Figure 3) regarding
a web page and the user can also view the previous reviews
belonging to the corresponding web page. Finally, the Statistics
tab in the extension home page illustrates statistics regarding
the particular URL (Figure 4).
The reputation server has been developed using PHP as
a web service exposing an endpoint. The extension interacts
with the server using this endpoint. It is also equipped with a
MySQL database where the reputation score, corresponding
reviews and statistics for each URL are stored. When the
Fig. 2: Extension Home Page.
Fig. 3: Extension Review Interface.
reputation server receives a request regarding the score of a
URL, it first checks in the database if there is any relevant data
regarding that URL. If found, data is returned to the extension.
If no data is found, a new entry is created for the URL with
a base reputation score. Once the server receives reviews of
a URL, it stores the review in the corresponding entry of the
URL in the database. Finally, when the server receives a rating
of a URL, it utilizes the reputation computation engine to
update its reputation.
The reputation computation engine employs a multinomial
Bayesian reputation system to calculate the reputation of a
URL. A Bayesian multinomial system mainly is based on
Dirichlet probability distribution [15] and allows a reputation
score to be expressed using different rating levels. In our
development we have employed a five scale rating where 1
corresponds to a deceptive content and 5 corresponds to a
trustworthy content. A multinomial system also requires a base
score for each rating level, especially useful when a URL has
no previous score recorded in the database. Our employed
Fig. 4: Extension Score Percentage.
rating level with the base score is presented in Table I.
Level Verbal tag Base score
L1 Deceptive 0.2
L2 Untrustworthy 0.2
L3 Average 0.2
L4 Trustworthy 0.2
L5 Very Trustworthy 0.2
TABLE I: Rating levels with base rates.
Based on the base score in Table I, the reputation engine
uses the formula 2of Section II to calculate a point estimate
reputation score normalized in the scale of 1.0 to 5.0. The
engine then updates the reputation score of the corresponding
URL in the database. Next, the score is transformed to a
meaningful textual tag (Figure 2) using a threshold range.
Both the reputation score and textual tag along with the
review and statistics for the corresponding URL are sent
back to the extension. It is to be noted that if a URL has
no associated reputation score, an average score of 3.0 is
assigned to bootstrap the system. For the first few ratings, each
score changes slightly as the system leverages the Bayesian
Multinomial System.
In this section, we present the user study that has been car-
ried out to test the usability and applicability of CrowdsouRS
along with a discussion of subjects, the tasks they have been
required to carry out, the questionnaires they have answered
and the corresponding result.
For the study, we have prepared a dataset of more than
600 URLs including some most popular, less popular and
hoax web sites involving Facebook, Google, ESPNcricinfo,
Prothom-Alo, Bdnews24,, and
40 subjects have been recruited for the user study via word
of mouth and personal references. The majority (95%) of them
are students aged between 18-40 and are active Internet users.
Among them, around 80% are computer science major.
Before starting, we have explained CrowdsouRS to the
participants and the problem we are trying to tackle using this
At first, the participants have been asked to fill up a
form containing several demographic questions1such as age,
occupations, etc. and other questions involving their internet
consciousness and safety awareness. Then, they have been
asked to install the extension in their browsers and randomly
visit any page from the dataset. The extension would show the
reputation score of the corresponding web page as discussed
before. If any user wishes, she can also rate a web page from
their own evaluation. Once they visit a sufficient number of
pages, they are asked to take part in a survey by filling out a
post questionnaire 2.
The purpose of the post questionnaire is mainly to collect
qualitative user experience on the usability of our system, and
to find out their thoughts on how this can be helpful, what
features to add in terms of reputation of a website.
From the two questionnaires we have found that 51%
participants have been fooled by misleading contents in their
experience whereas around 29% have stated that they might
be a victim of deceptive or misleading contents. We have
also found that 29.3% and 19.5% of participants have seen
deceptive online contents often and very often, respectively.
Figure 5illustrates the view of participants regarding
CrowdsouRS. As evident from the figures, most of the par-
ticipants (around 70%) believe that CrowdsouRS can be an
effective tool to fight against deceptive misleading online
Strongly Disagree Disagree Neutral Agre e Strongly A gree
Fig. 5: Influence of CrowdsouRS
The result of average reputation scores for different web-
sites as generated by CrowdsouRS utilizing the ratings pro-
vided by the participants is presented in Figure 6. According
to this figure, more than 70% people assume that Facebook
and Google are very trustworthy while only less than 2%
people mark these web sites untrustworthy. However, when
we calculate the reputation score of, 75%
people rightfully indicate that this website contains deceptive
misleading contents. On the other hand, the reputation scores
1Pre Study Questionnaire:
2Post Study Questionnaire:
Facebook Google Prothom-Alo bdnews24
Very Trustworthy Trustworthy Average Untrustworthy Very Unrustworthy
Fig. 6: URL Ratings from User According to Input.
indicate that most of the people have tagged Prothom-Alo,
BdNews24, ESPNcricinfo as neutral websites.
This shows that CrowdsouRS has been able to identify
deceptive contents which validates our assumption that the
wisdom of crowd can be leveraged for identifying decep-
tive contents. We have also evaluated the experience and
perception of users regarding CrowdsouRS. The evaluation
result is presented Figure 7. It is evident from the result
that the user experience has been mostly positive. Among
all users, around 41.5% participants have been satisfied and
22% of the participants have been highly satisfied while 29.3%
participants remain neutral.
The success of any crowd-sourced system depends on
its wide-scale adoption with a large amount of user input.
In our study, we have had a limited number of users with
the majority of our participants have been from the campus-
oriented neighborhood (95% were students, and others were
recent ex-students). The age and technically knowledgeable
inclination of our members may give a skewed vision of the
general convenience of our instrument. A wide-scale study
involving users from all spheres of life would be ideal to
present a more accurate representation of the wisdom of crowd.
We plan to address this issue by carrying out a larger user study
in the future with a more comprehensive dataset.
There are a good number of researches within the domain
of trust and reputation systems and their management. In this
section, we present a few influential related works.
Noorian and Ulieru [19] introduced a multidimensional
framework to compare various types of trust and reputation
systems which could be used to justify the applicability and
the utility of those systems.
In [13], [15] Jøsang et al. presented a survey of trust and
reputation systems for web services. Their focus was to give
an overview of existing systems as well to briefly discuss how
reputation scores in different systems were computed using
different mechanisms such as simple summation or average,
Bayesian calculation, belief model, fuzzy model, etc.
Highly Un satisfied Unsatisf ied Neutral Satisfie d Highly Sa tisfied
Fig. 7: Satisfaction with the Reliability of System.
In a similar vein, the background, the current state and
the future of Internet-mediated trust and reputation systems
had been presented in [20]. The authors attempted to show
that the multinomial Bayesian system with different levels
of reputation score provides great flexibility and usability.
Google’s PageRank and the slashdot model were also shortly
Whitby and Jøsang proposed a system to filter out unfair
ratings from the online marketplace[21]. In their article, they
showed how a Bayesian reputation system could be used to
filter product ratings.
A game theoretical model for reputation management in
online auction sites had been presented in [16] where it was
discussed how a Bayesian model averaging could be used in
a reputation system to integrate different information sources.
In [22], Kariyawasam and Weresinghe attempted to present
a reputation model which was devised to measure the reputa-
tions of reviews in a web-based review system. Their model is
based on the credibility of sources expressed as expertise and
Many existing reputations systems fail to effectively cal-
culate a reputation score when some buyers intentionally give
unfair ratings to a target. To address this problem, Dellarocas
and Chrysanthos proposed a solution to reduce the negative
effects of such fraudulent behaviour [23].
Web of Trust (WoT) [24] is an online service provider
that calculates the reputation score of different websites. In
a similar fashion like CrowdsouRS, it provides a browser
extension that effectively measures if a website is secure
and privacy friendly. Even though their mechanism is closely
related to our system, there are two major differences. Firstly,
their system does not identify any deceptive contents as their
goal is different. Secondly, the reputation score in WoT is
domain specific, meaning all pages under a domain would have
a similar score, where our system is URL specific, generating
different scores for each web page.
SmartNotes is also another crowdsourced system by Shar-
ifi et al.[25] which aims to detect security threats such as
Internet scams and deceitful sales of low standard products.
Their system mainly analyzes user feedback applying machine
learning techniques and natural language processing. However,
they do not utilize any reputation system and they do not aim
to identify any type of fake news.
The web has been one of the truly transformative technolo-
gies having a huge impact in all parts of our life every single
day. Unfortunately, the open nature of web has also fueled
the proliferation of deceptive misleading contents such as fake
news. It has increasingly become a major concern which needs
to be tackled with urgency.
In this paper, we have presented CrowdsouRS a crowd-
sourced reputation system that leverages the wisdom of the
crowd to identify deceptive online contents. It allows a user
to rate any web page via a browser extension. The individual
rating of each user is aggregated by a back-end reputation
server, based on multinomial Bayesian system, to calculate a
reputation score for a web page. The reputation score is then
displayed via the extension as a visual cue to aid other users
identify any deceptive content. A user study carried out with
a number of users highlights the tool to be effective as well
as usable.
In future, we plan to move further with a major upgrade of
the extension with an addition of machine learning approach
with the a large dataset to make it more accurate and provide
extensions for other popular browsers. We hope our approach
will pave down a novel way for tackling deceptive online
[1] S. Olivia, “Facebook staff mount secret push to tackle fake news,
reports say,” in the Guardian, 15 November 2016. [Online]. Avail-
fake-news-us- election-news- feed-algorithm [Accessed: September 09,
[2] A. Jamieson and O. Solon, “Facebook to begin flagging fake news in
response to mounting criticism,” in the Guardian, 15 December 2016.
[Online]. Available:
15/facebook-flag- fake-news-fact- check [Accessed: September 09,
[3] S. Levin, “Facebook promised to tackle fake news. but the
evidence shows it’s not working,” in the Guardian, 16 May 2017.
[Online]. Available:
may/16/facebook-fake-news-tools-not- working [Accessed: September
09, 2017].
[4] M. I. Daisuke Wakabayashi, “In race against fake news,
google and facebook stroll to the starting line - the new
york times,” in the New York Times, 25 January 2017.
[Online]. Available:
google-facebook- fake-news.html?mcubz=1 [Accessed: September 09,
[5] L. Leong, “Fighting fake news: how google, facebook
and others are trying to stop it — techradar,
in the Tech Radar, 25 May 2017. [Online]. Avail-
facebook-and- more-are- working-to-stop-it [Accessed: September 09,
[6] S. Gibbs, “Google to display fact-checking labels to
show if news is true or false — technology — the
guardian,” in the Guardian, 7 April 2017. [Online]. Avail-
to-display- fact-checking- labels-to- show-if-news- is-true- or-false [Ac-
cessed: September 09, 2017].
[7] S. Correspondent, “Kaler kantho resorted to yellow journalism — the
daily star,” in the Kaler-Kantho, 21 April 2011. [Online]. Available: [Accessed: September
09, 2017].
[8] S. Wazed, “Bangladesh fights malicious facebook postings,
online hate,” in the Diplomat, 12 December 2016.
[Online]. Available:
malicious-facebook- postings-online- hate/ [Accessed: September 09,
[9] J. Lorenz, H. Rauhut, F. Schweitzer, and D. Helbing, “How social
influence can undermine the wisdom of crowd effect,Proceedings of
the National Academy of Sciences, vol. 108, no. 22, pp. 9020–9025,
[10] W. Pan, Y. Altshuler, and A. Pentland, “Decoding social influence and
the wisdom of the crowd in financial trading network,” in Privacy,
Security, Risk and Trust (PASSAT), 2012 International Conference on
and 2012 International Confernece on Social Computing (SocialCom).
IEEE, 2012, pp. 203–209.
[11] A. Jøsang, E. Gray, and M. Kinateder, “Simplification and analysis
of transitive trust networks,Web Intelligence and Agent Systems: An
International Journal, vol. 4, no. 2, pp. 139–161, 2006.
[12] C. Dictionary, “Reputation meaning in the cambridge english
dictionary,” in the Cambridge Dictionary. [Online]. Available: [Accessed:
September 09, 2017].
[13] A. Jøsang, R. Ismail, and C. Boyd, “A survey of trust and reputation
systems for online service provision,” Decision support systems, vol. 43,
no. 2, pp. 618–644, 2007.
[14] A. Jøsang and W. Quattrociocchi, “Advanced features in bayesian
reputation systems,” in International Conference on Trust, Privacy and
Security in Digital Business. Springer, 2009, pp. 105–114.
[15] A. Jøsang and J. Haller, “Dirichlet reputation systems,” in Availability,
Reliability and Security, 2007. ARES 2007. The Second International
Conference on. IEEE, 2007, pp. 112–119.
[16] P. Nurmi, “A bayesian framework for online reputation systems,” in
Telecommunications, 2006. AICT-ICIW’06. International Conference on
Internet and Web Applications and Services/Advanced International
Conference on. IEEE, 2006, pp. 121–121.
[17] Y. Wang and J. Vassileva, “Bayesian network-based trust model,” in
Proceedings IEEE/WIC International Conference on Web Intelligence
(WI 2003), Oct 2003, pp. 372–378.
[18] H. T. Nguyen, W. Zhao, and J. Yang, “A trust and reputation model
based on bayesian network for web services,” in Web Services (ICWS),
2010 IEEE International Conference on. IEEE, 2010, pp. 251–258.
[19] Z. Noorian and M. Ulieru, “The state of the art in trust and reputation
systems: a framework for comparison,Journal of theoretical and
applied electronic commerce research, vol. 5, no. 2, pp. 97–117, 2010.
[20] A. Jøsang, “Trust and reputation systems,” in Foundations of security
analysis and design IV. Springer, 2007, pp. 209–245.
[21] A. Whitby, A. Jøsang, and J. Indulska, “Filtering out unfair ratings in
bayesian reputation systems,” in Proc. 7th Int. Workshop on Trust in
Agent Societies, vol. 6, 2004, pp. 106–117.
[22] S. Kariyawasam and R. Weerasinghe, “A unified framework for reviewer
reputation estimation in online review systems,” in Industrial and
Information Systems (ICIIS), 2015 IEEE 10th International Conference
on. IEEE, 2015, pp. 407–412.
[23] C. Dellarocas, “Immunizing online reputation reporting systems against
unfair ratings and discriminatory behavior,” in Proceedings of the
2Nd ACM Conference on Electronic Commerce, ser. EC ’00. New
York, NY, USA: ACM, 2000, pp. 150–157. [Online]. Available:
[24] T. A.-K. Sami Tolvanen, “Safe web search and secure browsing
— wot (web of trust),” July 1 2006. [Online]. Available: https:
// [Accessed: September 09, 2017].
[25] M. Sharifi, E. Fink, and J. G. Carbonell, “Smartnotes: Application of
crowdsourcing to the detection of web threats,” in Systems, Man, and
Cybernetics (SMC), 2011 IEEE International Conference on. IEEE,
2011, pp. 1346–1350.
... This in turn allows organizations to predict the performance of an extant or upcoming product in its intended market effectively because the opinions of consumers are reflected more strongly in such products provided crowdsourcing is used to design or modify the products in question. Some progress in vetting the authenticity and correctness of media content using crowdsourced opinions has already been made by researchers, through implementation of crowd-sourced online reputation systems [11]. The effect of social interactions in design and maintenance of content-centric networks has also been examined in recent years [12]. ...
Conference Paper
Crowdsourcing is one of the newest concepts in technology, with diverse applications. The current research focuses on using the concept of crowdsourcing to generate media content. Opinions sourced from the intended target audience are to be used for positively enhancing the impact of the media on its audience by catering to their demands, determined through multiple short surveys. The proposed mathematical estimation model is presented along with the simulation results obtained.
Conference Paper
Full-text available
Fake news is a problem for society as some people trust the message and react accordingly, propagating the lie. The consequences vary from a simple laugh to a death sentence. Research has focused on distinguishing between fake and fact through machine learning techniques. Most literature reviews and mappings focus on techniques that identify the fake over the message content. In a complementary way, we investigate fake news detection approaches based on the senders’ behavior on social media. We present a conceptual framework that deep dives into different aspects of the main categories of persuader, clarifier, and gullible actors. We also consider automatic and manual profiles. The results provide an overview of existing works on profile trustworthiness in fake news spread and shed light on possible research directions.
Full-text available
This systematic mapping literature aims to identify current research and directions for future studies in terms of combating cyber propaganda in the social media, which is used by both human effort and technological approaches (socio-technical) for mitigation. Out of 5176 retrieved articles, only 98 of them were selected for primary studies; classified based on research artifacts, mitigation effort, and the social media platforms involved in the research. The search was conducted using selected databases and applying selection criteria set for this research. Through the analysis, important research trends were identified based on human effort and technological approaches in mitigating and combating the cyber-propaganda issues. The authors also identified various mitigation socio-technical approaches such as identification, detection, image recognition, prediction, truth discovery and comprehension of rumours flow. The study also highlights areas for further improvements, to complement the performances of existing techniques. Besides, the study provides a brief review of cyber propaganda detection using classification techniques. Hence, it has set forth applicable research focus on the areas dealing with the mitigation of risk borne by cyber propaganda in the social media.
Conference Paper
Full-text available
We describe a crowdsourcing system, called SmartNotes, which detects security threats related to web browsing, such as Internet scams, deceptive sales of substandard products, and websites with intentionally misleading information. It combines automatically collected data about websites with user votes and comments, and uses them to identify potential threats. We have implemented it as a browser extension, which is available for free public use.
Conference Paper
Individuals and organizations widely use product reviews for their decision making. But most of the review systems pay less attention on verifying the identity of reviewers and the content of reviews. This leads to the presence of uninformative, biased or even false information. Also some people attempt to control the system by opinion spamming such as writing fake reviews to promote or demote some target products. If reviewer reputation can be measured, the consumers can find reviews that reflect genuine user experiences and opinions, depending on the reviews from reputed reviewers. This paper presents a reputation model devised to measure reviewer reputation in online review systems based on the source credibility which is defined as expertise and trustworthiness. Our experimental results show that the proposed reputation model is effective in measuring reputation. The model outperforms existing methods in measuring the reputation of reviewers. The model is simple and applicable to most of the online review systems.
Conference Paper
Bayesian reputation systems are quite flexible and can relatively easily be adapted to different types of applications and environments. The purpose of this paper is to provide a concise overview of the rich set of features that characterizes Bayesian reputation systems. In particular we demonstrate the importance of base rates during bootstrapping, for handling rating scarcity and for expressing long term trends.
Conference Paper
In this paper, we study roles of social mechanisms in a financial system. Our data come from a novel on-line foreign exchange trading brokerage for individual investors, which also allows investors to form social network ties between each other and copy others' trades. From the dataset, we analyze the dynamics of this connected social influence systems in decision making processes. We discover that generally social trades outperform individual trades, but the social reputation of the top traders is not completely determined by their performance due to social feedback even when users are betting their own money. We also find that social influence plays a significant role in users' trades, especially decisions during periods of uncertainty. We report evidences suggesting that the dynamics of social influence contribute to market overreaction.
Foundations of Security Analysis and Design.- Maude-NPA: Cryptographic Protocol Analysis Modulo Equational Properties.- An Introduction to Certificate Translation.- Federated Identity Management.- Electronic Voting in the Netherlands: From Early Adoption to Early Abolishment.- Logic in Access Control (Tutorial Notes).- The Open-Source Fixed-Point Model Checker for Symbolic Analysis of Security Protocols.- Verification of Concurrent Programs with Chalice.- Certified Static Analysis by Abstract Interpretation.- Resource Usage Analysis and Its Application to Resource Certification.- Analysis of Security Threats, Requirements, Technologies and Standards in Wireless Sensor Networks.
Trust and reputation systems represent a significant trend in decision support for Internet mediated service provision. The basic idea is to let parties rate each other, for example after the completion of a transaction, and use the aggregated ratings about a given party to derive a trust or reputation score, which can assist other parties in deciding whether or not to transact with that party in the future. A natural side effect is that it also provides an incentive for good behaviour, and therefore tends to have a positive effect on market quality. Reputation systems can be called collaborative sanctioning systems to reflect their collaborative nature, and are related to collaborative filtering systems. Reputation systems are already being used in successful commercial online applications. There is also a rapidly growing literature around trust and reputation systems, but unfortunately this activity is not very coherent. The purpose of this article is to give an overview of existing and proposed systems that can be used to derive measures of trust and reputation for Internet transactions, to analyse the current trends and developments in this area, and to propose a research agenda for trust and reputation systems.
Conference Paper
Trust and reputation for web services emerges as an important research issue in web service selection. Current web service trust models either do not integrate different important sources of trust (subjective and objective for example), or do not focus on satisfying different user's requirements about different quality of service (QoS) attributes such as performance, availability etc. In this paper, we propose a Bayesian network trust and reputation model for web services that can overcome such limitations by considering several factors when assessing web services' trust: direct opinion from the truster, user rating (subjective view) and QoS monitoring information (objective view). Our comprehensive approach also addresses the problems of users' preferences and multiple QoS-based trust by specifying different conditions for the Bayesian network and targets at building a reasonable credibility model for the raters of web services.
Conference Paper
Reputation systems can be used in online markets and communities in order to stimulate quality and good be- haviour as well as to sanction poor quality and bad behaviour. The basic idea is to have a mechanism for rating services on various aspects, and a way of computing reputation scores based on the ratings from many different parties. By making the reputation scores public, such systems can assist parties in deciding whether or not to use a particular service. Reputation systems represent soft security mechanisms for social control. This article presents a type of reputation system based on the Dirichlet probability distribution which is a multinomial Bayesian probability distribution. Dirichlet reputation systems represent a generalisation of the binomial Beta reputation system. The multinomial aspect of Dirichlet reputation systems means that any set of discrete rating levels can be defined. This provides great flexibility and usability, as well as a sound basis for designing reputation systems.
Conference Paper
There are currently very few practical methods for assessing the qual- ity of resources or the reliability of other entities in the o nline environment. This makes it difficult to make decisions about which resources ca n be relied upon and which entities it is safe to interact with. Trust and repu tation systems are aimed at solving this problem by enabling service consumers to reliably assess the quality of services and the reliability of entities befo re they decide to use a particular service or to interact with or depend on a given entity. Such systems should also allow serious service providers and online players to correctly rep- resent the reliability of themselves and the quality of thei r services. In the case of reputation systems, the basic idea is to let parties rate e ach other, for exam- ple after the completion of a transaction, and use the aggregated ratings about a given party to derive its reputation score. In the case of tru st systems, the basic idea is to analyse and combine paths and networks of trust relationships in order to derive measures of trustworthiness of specific nodes. Rep utation scores and trust measures can assist other parties in deciding whether or not to transact with a given party in the future, and whether it is safe to depend on a given resource or entity. This represents an incentive for good behaviour and for offering reliable resources, which thereby tends to have a positive effect on the quality of online markets and communities. This chapter describes the background, current status and future trend of online trust and reputation systems.