See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/227859089

A Concise Study of Web Filtering

Article · January 2008

CITATIONS READS

2 2,590

2 authors, including:

M. Tariq Banday
University of Kashmir
140 PUBLICATIONS   424 CITATIONS   

SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Book captioned "Cryptographic Security Solutions for the Internet of Things" to be published by IGI Global, USA View project

inderscience paper View project

All content following this page was uploaded by M. Tariq Banday on 23 May 2014.

The user has requested enhancement of the downloaded file.

 Working Papers on Information Systems ISSN 1535-6078

A Concise Study of Web Filtering

M. Tariq Banday
University of Kashmir, India
Nisar A. Shah
University of Kashmir, India

Abstract
Cybercriminals are constantly developing techniques to infect computers by embedding
malicious code on innocent websites and luring victims to them. To prevent data loss in a
mobile connected world, corporations are employing a variety of techniques. These include
filters, anti-virus software, encryption and firewalls, access control, written policies and
improved employee training. This paper conducts a concise study of web filtering vis-à-vis
their installed positions, deployment layers, employed filter technologies and comparison
between Web Filters that are in place in Canada, United Kingdom, and China.

Keywords: Web Filtering, Filter Deployment, Filter Operating Layers, Rating Filters,
Blacklisting, Keyword Matching, Dynamic Filtering.

Permanent URL: http://sprouts.aisnet.org/10-31

Copyright: Creative Commons Attribution-Noncommercial-No Derivative Works License

Reference: Banday M.T., Shah N.A. (2010). "A Concise Study of Web Filtering," . Sprouts:
Working Papers on Information Systems, 10(31). http://sprouts.aisnet.org/10-31

Sprouts - http://sprouts.aisnet.org/10-31
 A Concise Study of Web Filtering

M. Tariq Banday, Nisar. A. Shah

P.G. Department of Electronics and Instrumentation Technology,
University of Kashmir, Srinagar - 6, India
E-mail: [email protected]

Abstract

Cybercriminals are constantly developing techniques to infect computers by embedding malicious

code on innocent websites and luring victims to them. To prevent data loss in a mobile connected
world, corporations are employing a variety of techniques. These include filters, anti-virus software,
encryption and firewalls, access control, written policies and improved employee training. This paper
conducts a concise study of web filtering vis-à-vis their installed positions, deployment layers,
employed filter technologies and comparison between Web Filters that are in place in Canada, United
Kingdom, and China.

Keywords

Web Filtering, Filter Deployment, Filter Operating Layers, Rating Filters, Blacklisting, Keyword
Matching, Dynamic Filtering.

Introduction

Web filtering is a class of content filtering techniques [1] used by corporations and home users as a
part of Internet firewall to determine whether incoming data is harmful to the network or outgoing
data includes any intellectual property. The filter checks every Web page against a set of predefined
rules and blocks harmful and objectionable data like pornographic material, spyware, viruses, etc.
from entering the network or the home computer.

Web Filtering guarantees manageable Internet access by reducing the unnecessary use of network
resources, increasing work productivity, decreasing risks of Internet abuse, and decreasing security
and legal risks.

More than forty Western and non-Western countries including Saudi Arabia, Iran, Norway, Sweden,
Denmark, UK, and Netherlands are using Web filters to block Websites considered to be
inappropriate.

Sprouts - http://sprouts.aisnet.org/10-31
 Banday, M.T. and Shah, N.A.
 

Filter Deployment

A Web Filter can be installed at various places in the Network and may operate at various levels of
the OSI Model as depicted in figure 1. The legends1 through 5 denote the place of filter as explained
in the below paragraphs. The customization options, Performance of the filter and Security provided
depend greatly on the place of the deployment [2, 3].

1. At National/Country Level: The Filter is deployed between the national Internet

Backbone and the country network. Several nations including China and Saudi Arabia
have implemented filters at National Level. Filter Configuration is wholly determined by
the Governmental. Users have no control over filter customization, its performance and
thus security provided by the filter is determined by the policy of the Government.

2. At Organizational Level: The Filter is deployed between the Organizational Network and
the Internet Gateway. All users of this Gateway Server e.g. all employees of the
Organization are provided filtered Internet Content. The KU Gateway installed at
http://192.168.81.251:8090/corporate/servlet/CyberoamHTTPClient is an example of
Organizational Level Filtering. The filter is customizable by the Organizations Web
Administrators keeping into consideration the Organizational policy regarding what is
appropriate and what is not for the organization. Organizations besides providing filtered
content also can limit the durations of use of the Internet.

3. At Internet Server Provider: The Filter is installed at the ISP Gateway and provides
filtered content to all its customers. Informal Government pressures in Canada and UK
led major ISPs to voluntarily institute filtering to block inadequate access to child
pornography and child abuse material. Courts in France, Belgium and Germany have
ordered ISPs to block hate speeches and the illegal peer to peer file shearing of copyright
protection material. As per our knowledge no ISP in India provides filtered content to its
customers.

4. At Individual Level: The Filter is installed at the local computer or workstation. The
Filter may be part of a Firewall, Antivirus package, or through some other similar system
like Content Advisor, Parental Control, etc.

5. At Third-Party: The Filtering service is provided by a trusted third party vendor through
its Security Operation Canters (SOCs). The customers send their Web Traffic through
these SOCs by proxying. ScanSafe and WebSense are the examples of such Third-Party

A Concise Study of Web Filtering - Sprouts 2010

Page 2 of 11 
 
Sprouts - http://sprouts.aisnet.org/10-31
 Banday, M.T. and Shah, N.A.
 

Vendors. Although suitable for all kinds of organizations and users, this service is limited
to small and medium organizations but offers filtering at any level of an organization.

Figure 1: Filter Deployment

Filter Operating Layers

The Filter installed at various places in the network may operate at either layer 3 called Network
Layer or layer 4 called Transport Layer or at Layer 7 called Application Layer of the OSI
Networking Model. Filters installed at Layers 3 and 4 are referred to as Network Layer Filters and
those installed at Layer 7 are called Application Layer Filters.

1. Layer 3 of the OSI model is responsible for logical addressing and routing of data using
protocols like IP. The packet contains source and destination addresses which can be used to
block the transmission of the packet based on some defined rules of the filter.

A Concise Study of Web Filtering - Sprouts 2010

Page 3 of 11 
 
Sprouts - http://sprouts.aisnet.org/10-31
 Banday, M.T. and Shah, N.A.
 

2. Layer 4 of OSI model is responsible for formatting and transporting data using protocols like
TCP and UDP. The packet at this layer contains source and destination addresses besides
containing information about the type of network traffic thus enabling blockage of traffic
from certain address meant for a particular application.

3. Layer 7 of the OSI model is responsible for data analysis before sending it to a particular
application. At this layer packets are assembled and thus inspection of the data arriving for a
particular application can be undertaken by performing deep inspection for the content
filtering. Application Proxy firewall operates at this layer of the OSI model.

• Enables user to accept to access the network. Provides user Interfaces and
Application Layer support to services like email, WWW, FTP, remote Login, etc. Gateway
Layer 7 Operates at this Layer. Addressing varies on the basis of service e.g. email
addresses, URL Addresses, FTP connection addresss, etc.

Presentation Layer • Handles the Systax and Sementics of the information exchanged between
two applications. Provides translation of Syntax between different systems,
Layer 6 encryption, decryption and compression.

• Acts as a network dialog controller. Establishes, maintains and

Session Layer synchronizes the interaction between two systems.. Provides Dialog Control
Layer 5 and Dialog Seperation Services.

• Responsible for Process to Process Delivery. Provides QOS through

Connection Oriented and Connectionless transmission of IP datagrams in
Transport Layer Transport layer packets. Uses Protocols Like TCP,UDP, SPX. Does Port
Layer 4 Addressing, Segementation and reassembly, Connection control Flow
Control and error Control.
• Responsible for End to End Transmission of datagrams across an
Network Layer Internetwork including Fragmenting , LOgical Addressing and
Routing.Uses Protocols Like IP, IPX, NetBEUI. Router Operate at this
Layer 3 Layer. Messaging unit is Datagram which includes IP addresses,
identifying computers on Internetwork.
• Responsible for packet transmission betwween two systems on LAN. Does
Datalink Layer framing, Flow Control, error Control, Access Control and Physical
Addressing. Messaging Unit is Frame and addressing through Hardware
Layer 2 Addresses (NIC Addresses). Protocols Like Ethernet, Token Ring, etc. are
used. Bridges and Switches Operate at this Layer.

Physical Layer • Responsible for Transmission and reception of bits of data on /from the
Medium, Signalling, Method and Cabeling, Representation of Bits, data
Layer 1 Rate and Synchronization.

Figure 2: OSI Model

Filtering Techniques

Web Filtering techniques [4, 5, 6, 7] vary on the basis of their workings and the data they
work upon. Figure 3 shows four possible filtering techniques namely rating based,

A Concise Study of Web Filtering - Sprouts 2010

Page 4 of 11 
 
Sprouts - http://sprouts.aisnet.org/10-31
 Bandaay, M.T. and Shah,
S N.A.
 

blacklissting, keywoord matchinng and dynaamic that work

w upon diifferent infoormation associated
with thee web conteent. All of these
t filterinng techniqu
ues can be used
u at the aapplication layer of
the OSII model butt keyword matching
m annd dynamicc filtering can
c effectiveely be used
d only at
applicattion layer.

Filtering 
T
Technique es

Keyword 
Rating Blacklistin
ng Dynamic
M
Matching

Figure 3:
3 Filtering Teechniques

1. Ratiing: Worldd Wide Weeb Consortiium (W3C)) has introduced a laabelling sysstem
nam
med Platform
m for Interrnet Contennt Selection
n (PICS) thhat defines a platform
m for
creaation of content labellling system
m. It enablees the authhors of the Web pagees to
incllude labels also
a called metadata
m thhat describess the contennt of the pagge. On the basis
b
of this metadatta third parrty rating auuthorities lik
ke Internet Content Raating Autho
ority
(ICR
RA) rates thhe website on
o the basiss of presencce or absencce of certainn elements in
i it.
Thee ratings innclude ratinngs for Nuddity, Sexuaal Content, Weapon U
Use, Drug Use,
U
Vioolence, etc. A file is theen generatedd containing
g ratings and the label tthat is linkeed to
the domain of the
t Websitee. Web brow
wsers like In
nternet Explorer, Safarri, and Netsccape
incllude a Conttent Advisoory mechannism that heelps the useers to regullate the con
ntent
theyy want to bllock.

This rating system

s is not
n regulateed as somee Web authhors in ordeer to evadee the
poossibility off their Webb content beeing blockeed do not innclude metaadata or include
inncorporate metadata
m in their Web content. Th
hus Content Advisors ddo not proviide a
fooolproof sollution and should
s be included as an additionnal line of ddefence agaainst
poornographyy.

2. Blaccklisting: This
T techniqque uses a URL
U catego
orization daatabase wheere URLs have
h
beenn mapped to different categories according
a to
o their conttent. The W
Web filter po
olicy

A Concise Study
S of Web F
Filtering - Sprrouts 2010
Pagge 5 of 11 
 
Sprouts - http://sprouts.aisnet.org/10-31
 Banday, M.T. and Shah, N.A.
 

decides which categories to pass and which to block. The URLs belonging to the
categories to be blocked constitutes the blacklist. The filter compares the requested
URL against the blacklist and allows or denies this request accordingly. It is also
possible to blacklist on the basis of IP address and Domain name besides URL
Blocking at IP address level permits blocking of all domains hosted on the
corresponding Web Server. Blocking at Domain name level blocks the entire domain.

The advantages of this method are speed and efficiency because the filter based on the
blacklist has not to read the page before blocking or allowing.

Its disadvantages are the difficulties faced to create, and update the URL database as it is
labour-intensive and requires human reviewers. Human reviewers nowadays have been
replaced by automated filtering where a spider program automatically does
categorization.

3. Keyword Matching: This type of filtering works by inspecting the web traffic for
certain offensive words like ‘teen’, ‘sex’, ‘breast’ etc. and phrases, comparing them
with its set of words and phrases to determine whether to allow or deny its access.
Keyword matching filters is purely text-based methods. Keyword filtering is fast but
over-blocking errors may be produced by this type of filter if the words labelled as
offensive appear in legitimate web pages like sexton, breast cancer, etc. More precise
content analysis methods can be used to reduce over-blocking but at the same time
processing time will increase. Further, the efficiency of this filter for filtering
pornography content is less because pornographic material often includes hefty data
in pictorial and video formats.

4. Dynamic Filtering: These filters use various statistical machine learning methods like
Baysian, k-Nearest Neghibour, etc. to understand the semantic content of the
information to be filtered. They use multiple features; features from text (words like
‘sex’, ‘teen’, ‘gambling’, etc.), images (photographs in nudity), and possibly video
clips. For filtering images for pornography colour, shapes and skin are investigated by
algorithms like skin model, skin detection and regions of interest. Dynamic filters can
be trained and continue to learn more with use. Several dynamic filters with
reasonable efficiency to block pornography are available as commercial products.

A Concise Study of Web Filtering - Sprouts 2010

Page 6 of 11 
 
Sprouts - http://sprouts.aisnet.org/10-31
 Banday, M.T. and Shah, N.A.
 

Dynamic filters can also be used to construct and maintain blacklist categorization
database.

Dynamic filtering offering advantage of automated filtering and learning capabilities can be
designed to have higher efficiency but only at the cost of speed of operation making it
unsuitable to be used at places like ISP and Organizational gateways.

Comparison of Filters Installed at National Level in various Countries

Table 1 show below shows a comparison between Web Filters that are in place in Canada, United
Kingdom, China and proposed filter of Australia.

Australia Canada United Kingdom China

Legislating Mandatory Filtering at ISP Level
Yes
Yes No No Around 20 pieces of
legislation affects filtering
Voluntary/Industry Filtering at ISP Level
Yes Yes Yes
Perhaps Informal Government Informal Government Corporate Self-censorship
Pressure Pressure is Prevalent
Opt-Out Provision
No (Tier 1)
No No No
Yes (Tier 2)
Blacklist Filtering of Blocked URLs
Yes Yes Yes Yes
Purpose of Blacklist
Blocking inadequate access Blocking inadequate access
to child pornographic to child pornographic Blocking various types of
Unspecified
material with HTTP material with HTTP illegal content
protocol protocol
Type of Material Blacked
Political Content, Graphic
Child Pornography and Violence, Unapproved
Child Pornography Child Pornography
other illegal content news, Child Pornography
and other illegal content
Blacklist maintained by
Ministry of Industry and
ACMA
information; Centre
Australian
Cybertip.ca Internet Watch Foundation Propaganda Department;
Communications and
Ministry of Posts and
Media Authority
Telecommunications
IP Address Blocking
No No No Yes
Deep Packet Inspection
No Yes Yes Yes
Purpose of Deep packet Inspection
Traffic Shaping,
NA Traffic Shaping Traffic Shaping Dataveillance and
Surveillance
Other Heuristic Methods
Yes Yes Yes Yes

A Concise Study of Web Filtering - Sprouts 2010

Page 7 of 11 
 
Sprouts - http://sprouts.aisnet.org/10-31
 Banday, M.T. and Shah, N.A.
 

Australia Canada United Kingdom China

P2P
Perhaps Perhaps Content
Content infringement (in infringement (in
No Yes
negotiation with music negotiation with music
industry) industry)
Instant Messaging
No No No Yes
Scope Creep
Yes
Yes Legislation written with
Yes
Suicide sites, graphic standard vague and
Inevitable Suicide sites, pro-terrorism
terrorist beheading, pro- ambiguous clauses such as
sites, hate sites
terrorism sites, hate sites the ‘state security’
provision
Offence to Circumvent Filters
Yes Yes
Not an offence to use Not an offence to use
No No
circumvention devices such circumvention devices such
as proxy for other purposes as proxy for other purposes
Legislative Safeguards
No
Limited No
No Bill of Rights, Limited
Charter of human Rights The human rights
Constitutionally implied Europe Convention on
does not bind corporations instruments are of little
freedom of political human Rights; relevant
such as ISPs (No practical significance (eg.
communication very case law from European
legislation compelling Freedom of Expression is
limited in this content and Court of Human Rights
ISPs) not an individual right)
of little use as a safeguard
Market Safeguard
Potentially Potentially
No Voluntary initiative subject Voluntary initiative subject
None
Compulsory for ISPs to strong informal to strong informal
government pressure government pressure
Technical Safeguard
Potentially Potentially
Potentially
Depends where the filtering Depends where the filtering
Geographic region of
routers are placed (e.g. routers are placed (e.g.
No access and bandwidth
router located on the router located on the
capability affect ability to
backbone would affect all backbone would affect all
access material
ISPs) ISPs)

Table 1: Comparison between Web Filters that are in place in Canada, United Kingdom, China and proposed
filter of Australia [8]

It is apparent from the above comparison that no uniform criteria for filtering have been
adopted by the compared countries apart from child pornography, URL Blacklisting, Instant
Messaging, and Heuristic Methods. All other parameters for filtering vary from country to
country.

A Concise Study of Web Filtering - Sprouts 2010

Page 8 of 11 
 
Sprouts - http://sprouts.aisnet.org/10-31
 Banday, M.T. and Shah, N.A.
 

Effectiveness and Limitations

Filters vary widely in their performance, and there is a trade-off between failing to block
unauthorized content called “under-blocking” and erroneously blocking authorized content
called “over-blocking”. Filters that block a large percentage of unauthorized content also
block a sizable percentage of authorized content in error. Web filters can make two types of
errors namely false positive also called over blocking and false negative also called under
blocking. Over blocking blocks permissible websites, raising issues about freedom of speech
and legal issues clamming damage. Under blocking allows inappropriate websites to pass
through the filter reducing its efficiency.

Several research works have reported that accuracy tests against filters do not provide a
conclusive ranking about its efficiency. This is due to the fact that a filter may be highly
accurate but it may be inefficient if only a few users are able to bypass it. Further,
information on the Internet changes in a rapid and continuous manner forcing the filters to
update at the same rate. A highly accurate filter may thus prove to be inefficient if it does not
update itself with this change.

An ideal filter that neither produces false positive not false negative errors does not exist and
thus a balance between the two filtering errors is highly desired. It has been found that filters
are not efficient against those who manually exchange pornographic material. But filters
reduce the availability of prohibited content and thus serve at least its modest objective of
protecting innocent users against abuse and exposure to sensitive material.

Conclusion

The study of web filtering reveals that filtering is possible at various places using a variety of
filtering technologies which may operate at either network layer, transport layer or
application layer of the OSI model. Depending upon the required customization of the
filtering criteria, position of the filtering system is determined. Positions close to the main
backbone leave no or very less filtering customization option for the ISP or the user. No
specific filtering technique is having cent percent accuracy. The performance of a good filter
may deteriorate unless it is constantly upgraded and maintained. The country level filtering

A Concise Study of Web Filtering - Sprouts 2010

Page 9 of 11 
 
Sprouts - http://sprouts.aisnet.org/10-31
 Banday, M.T. and Shah, N.A.
 

mechanism does not adopt any universal criteria and instead filtering criteria is decided by its
respective governments.

Biographies 

M. Tariq Banday was born in 1969. He did his M. Sc. and M. Phil. Degrees
from the Department of Electronics, University of Kashmir, Srinagar, India in
1996 and 2008 respectively. He did advanced diploma course in computers and
qualified UGC NET examination in 1997 and 1998. At present he is working
as Assistant Professor in the Department of Electronics & Instrumentation Technology,
University of Kashmir, Srinagar, India. He has to his credit several research publications in
reputed journals and conference proceedings. He is a member of Computer Society of India,
International Association of Engineers and ACM. His current research interests include
Network Security, Internet Protocols and Network Architecture.

Nisar A. Shah was born in 1953. He did his M. Sc. and Ph. D. Degrees from
the department of Physics, University of Kashmir, Srinagar, India in 1976 and
1981 respectively. At present he is working as Professor in the Department of
Electronics & Instrumentation Technology, University of Kashmir. He has to
his credit about 150 research publications which have been published in national and
international journals of repute. He has supervised several research scholars in M. Phil. and
Ph. D. programs. His current research interests include Digital Signal Processing and
Network Security.

References

[1]. Jose Maria Gomez Hidalgo, Enrique Puertas Sanz, Francisco Carrero Garcia, Manuel De
Buenaga Rodriguez, (2009 ), “Chapter 7 Web Content Filtering”, In: Marvin V. Zelkowitz,
Editor(s), Advances in Computers, Elsevier, Vol. 76, “Social Networking and The Web”, pp.
257-306, ISSN 0065-2458, ISBN 9780123748119, DOI: 10.1016/S0065-2458(09)01007-9.

[2]. W.Ph. Stol, H.K.W. Kaspersen, J. Kerstens, E.R. Leukfeldt, and A.R. Lodder, (2009),
“Governmental filtering of websites: The Dutch case”, Computer law & security review, vol. 25,
pp. 251–262.

A Concise Study of Web Filtering - Sprouts 2010

Page 10 of 11 
 
Sprouts - http://sprouts.aisnet.org/10-31
 Banday, M.T. and Shah, N.A.
 

[3]. Deibert R. J., Palfrey J. G., Rohozinski R., Zittrain J. (2008), “Access Denied; the Practice and
Policy of Global Internet Filtering”. Cambridge, Mass: The Mitt Press; 2008.

[4]. Michael Chau and Hsinchun Chen, (2008), “A machine learning approach to web page filtering
using content and structure analysis, Decision Support Systems”, vol. 44 pp. 482–494.

[5]. K. V. Chandrinos, Ion Androutsopoulos, G. Paliouras and C. D. Spyropoulos, (2000),

“Automatic Web Rating: Filtering Obscene Content on the Web”, Lecture Notes in Computer
Science, Volume 1923/2000.

[6]. Anirudh Ramachandran, Nick Feamster and Santosh Vempala, (2007), “Filtering spam with
behavioural blacklisting”, proceedings of the 14th ACM conference on Computer and
communications security, Pages: 342 – 351.

[7]. Patrick Reynolds and Amin Vahdat, (2003), “Efficient peer-to-peer keyword searching”,
Proceedings of the ACM/IFIP/USENIX 2003 International Conference on Middleware, Pages:
21-40.

[8]. Alana Maurushat and Renee Watt, (2009), “Clean Feed: Australia’s Internet Filtering Proposal”,
University of New South Wales, Faculty of Law Research Series, paper 7, 2009.

A Concise Study of Web Filtering - Sprouts 2010

Page 11 of 11 
 
Sprouts - http://sprouts.aisnet.org/10-31
 Working Papers on Information Systems | ISSN 1535-6078

Editors:
Michel Avital, University of Amsterdam
Kevin Crowston, Syracuse University

Advisory Board: Editorial Board:

Kalle Lyytinen, Case Western Reserve University Margunn Aanestad, University of Oslo
Roger Clarke, Australian National University Steven Alter, University of San Francisco
Sue Conger, University of Dallas Egon Berghout, University of Groningen
Marco De Marco, Universita’ Cattolica di Milano Bo-Christer Bjork, Hanken School of Economics
Guy Fitzgerald, Brunel University Tony Bryant, Leeds Metropolitan University
Rudy Hirschheim, Louisiana State University Erran Carmel, American University
Blake Ives, University of Houston Kieran Conboy, National U. of Ireland Galway
Sirkka Jarvenpaa, University of Texas at Austin Jan Damsgaard, Copenhagen Business School
John King, University of Michigan Robert Davison, City University of Hong Kong
Rik Maes, University of Amsterdam Guido Dedene, Katholieke Universiteit Leuven
Dan Robey, Georgia State University Alan Dennis, Indiana University
Frantz Rowe, University of Nantes Brian Fitzgerald, University of Limerick
Detmar Straub, Georgia State University Ole Hanseth, University of Oslo
Richard T. Watson, University of Georgia Ola Henfridsson, Viktoria Institute
Ron Weber, Monash University Sid Huff, Victoria University of Wellington
Kwok Kee Wei, City University of Hong Kong Ard Huizing, University of Amsterdam
Lucas Introna, Lancaster University
Sponsors: Panos Ipeirotis, New York University
Association for Information Systems (AIS) Robert Mason, University of Washington
AIM John Mooney, Pepperdine University
itAIS Steve Sawyer, Pennsylvania State University
Addis Ababa University, Ethiopia Virpi Tuunainen, Helsinki School of Economics
American University, USA Francesco Virili, Universita' degli Studi di Cassino
Case Western Reserve University, USA
City University of Hong Kong, China Managing Editor:
Copenhagen Business School, Denmark Bas Smit, University of Amsterdam
Hanken School of Economics, Finland
Helsinki School of Economics, Finland Office:
Indiana University, USA Sprouts
Katholieke Universiteit Leuven, Belgium University of Amsterdam
Lancaster University, UK Roetersstraat 11, Room E 2.74
Leeds Metropolitan University, UK 1018 WB Amsterdam, Netherlands
National University of Ireland Galway, Ireland Email: [email protected]
New York University, USA
Pennsylvania State University, USA
Pepperdine University, USA
Syracuse University, USA
University of Amsterdam, Netherlands
University of Dallas, USA
University of Georgia, USA
University of Groningen, Netherlands
University of Limerick, Ireland
University of Oslo, Norway
University of San Francisco, USA
University of Washington, USA
Victoria University of Wellington, New Zealand
Viktoria Institute, Sweden

View publication stats