Knowledge base compound approach against phishing attacks using some parsing techniques

Jan Zizka (Eds) : CCSIT, SIPP, AISC, PDCTA - 2013
pp. 33–40, 2013. © CS & IT-CSCP 2013 DOI : 10.5121/csit.2013.3604
KNOWLEDGE BASE COMPOUND APPROACH
AGAINST PHISHING ATTACKS USING SOME
PARSING TECHNIQUES
Gaurav Kumar Tak1
, Gaurav Ojha2
, Udit Kumar3
1
School of Computer Science & Information Technology, Lovely Professional
University, Phagwara, Punjab – 144402, India
gauravtakswm@gmail.com
2,3
Department of Information Technology, Indian Institute of Information
Technology and Management, Gwalior – 474010, India
ABSTRACT
The increasing use of internet all over the world, be it in households or in corporate firms, has
led to an unprecedented rise in cyber-crimes. Amongst these the major chunk consists of
Internet attacks which are the most popular and common attacks are carried over the internet.
Generally phishing attacks, SSL attacks and some other hacking attacks are kept into this
category. Security against these attacks is the major issue of internet security in today’s
scenario where internet has very deep penetration. Internet has no doubt made our lives very
convenient. It has provided many facilities to us at penny’s cost. For instance it has made
communication lightning fast and that too at a very cheap cost. But internet can pose added
threats for those users who are not well versed in the ways of internet and unaware of the
security risks attached with it. Phishing Attacks, Nigerian Scam, Spam attacks, SSL attacks and
other hacking attacks are some of the most common and recent attacks to compromise the
privacy of the internet users. This paper discusses a Knowledge Base Compound approach
which is based on query operations and parsing techniques to counter these internet attacks
using the web browser itself. In this approach we propose to analyze the web URLs before
visiting the actual site, so as to provide security against web attacks mentioned above. This
approach employs various parsing operations and query processing which use many techniques
to detect the phishing attacks as well as other web attacks. The aforementioned approach is
completely based on operation through the browser and hence only affects the speed of
browsing. This approach also includes Crawling operation to detect the URL details to further
enhance the precision of detection of a compromised site. Using the proposed methodology, a
new browser can easily detects the phishing attacks, SSL attacks, and other hacking attacks.
With the use of this browser approach, we can easily achieve 96.94% security against phishing
as well as other web based attacks
KEYWORDS
Internet attacks, Phishing, SSL, Cyber-crime, Knowledge Base, and Parsing.
1. INTRODUCTION
On the World Wide Web, Cyber-crime is one of the major security issues, troubling internet
security. These crimes can be defined as immoral actions performed with the use of internet. They
include illegal access of data, illegal interception of data, eavesdropping of authorized data over

34 Computer Science & Information Technology (CS & IT)
an information technology infrastructure , data interference (which includes unauthorized
damaging, deletion, deterioration, alteration or suppression of computer data), Unethical access of
web services, Disturbance of social-peace, systems interference (interfering with the functioning
of a computer system by inputting, transferring, destroying, removing, deteriorating, altering or
suppressing computer data), misuse of devices, forgery (ID theft), and electronic fraud.[1][4]
A Knowledge Base is the modelling of previously occurred events in order to predict future
events by employing some artificial intelligence techniques [13]. It is a sort of database for
knowledge management, providing the means for the computerized collection, organization, and
retrieval of knowledge. They are basically artificial intelligent tools providing intelligent
decisions. Knowledge is obtained and represented using various knowledge representation
techniques rules, frames and scripts. The basic advantages offered by such system are
documentation of knowledge, intelligent decision support, self-learning, reasoning and
explanation. [14]
As we all know that phishing attack is a URL based attack which happens between the Internet
user and the browser, so our proposed methodology gives the new security layer between browser
and the User using the Knowledge Base and some parsing operations.
2. LITERATURE REVIEW
Commonly, anti-phishing tools use two major approaches for mitigating phishing sites. The first
approach is based on heuristics to check the host name and the URL for common spoofing
techniques. The second method lists out some blacklist phishing URLs. The heuristics approach is
not 100% accurate since it produces low false negatives (FN), i.e. a phishing site is mistakenly
judged as legitimate, which implies they do not correctly identify all phishing sites. The heuristics
often produce high false positives (FP), i.e. incorrectly identifying a legitimate site as fraudulent.
Blacklists have a high level of accuracy because they are constructed by paid experts who verify a
reported URL and add it to the blacklists if it is considered as a phishing website. [1][4][8]
Delayed password disclosure [15] is another new method to avoid phishing attacks. This is based
on the feedback generated by the interface as user enters the password; hence if the feedback
generated is not according to the authentic website an alarm is triggered.
Another method to create awareness amongst users against phishing is Trust bar construction
[16]. This method associates logos with the public key of the website being visited hence easing
the way of authentication of website. PassmarkTM
is a similar method currently being used by
Bank of America.
The detection and identification of phishing websites in real-time, particularly for e-
banking/payment gateway website, is a very complex and dynamic problem which involves many
factors and criteria. Many methods like improving site authenticity, one time passwords, having
separate login and transaction passwords, personalized e-mail communication, user education
about phishing are being implemented to prevent phishing attacks, but they don’t provide high
security.
3. PROPOSED METHODOLOGY
Here we propose a knowledge base approach against phishing attacks which also uses some
parsing techniques to detect the attack.

Computer Science & Information Technology (CS & IT) 35
3.1. Knowledge Bases
Our methodology uses some knowledge bases which are I, T, A, B & C. Knowledge Base Initial
or KBI stores the pattern and other detection methods of previously detected phishing attacks and
other web attacks. It validates the URL and also relates the URL with the previously detected
phishing attacks. If pattern of new URL matches with the previously stored Phishing attacks, then
it generates a phishing alert before visiting the URL. Knowledge Base Trusted or KBT maintains
all the trusted and secure URLs which are previously visited on the same browser. The user can
further manually add the frequently visited legitimate websites to this knowledge base for whom
he wishes not to carry out security checks every. Knowledge Base A defines all the URL-pattern
based phishing and SSL attacks which have detected previously by the browser till date. This
Knowledge Base is used before the operation of ‘Parser-1’.
Knowledge Base B stores the all information (like license year, rating of the domain, popularity
of the domain etc.) of the URLs which is previously visited and detected as the Phishing attacks.
Knowledge Base C stores the result of Fraud Check analysis of URLs and generates queries when
the URL is analyzed using fraud check with the previous history. Finally, Knowledge Base D is
responsible for maintaining the history of the all the URLs which are previously visited.
3.2. Parsers
Some Parsers are also used in the detection of URL based attack in the proposed methodology.
Parser 1 is used to detect the pattern based URL attacks. This parser provides the security against
phishing attacks as well as SSL attacks. It also analyzes the usage of some special character (like
‘-‘,’.’ etc.) in the URL to detect the attacks. This parser’s operation is based on the fact, that
phishing attackers use the some fraction of the actual legitimate URL so as to generate a close to
real phishing URL. Then we have Parser 2, which pulls out all the details of the website such as
license year, rating of the domain, popularity of the domain etc. when a URL is passed through it.
Using these details parser-B can declares if the URL is phishing website URL or a legitimate
website URL. This parser takes account of the fact that phishing URLs are newly registered one
with low rating and popularity. Hence if the URL is newly registered, then it can be a phishing
attack on any existing URLs.
Parser 3 performs an important step for security against the phishing attacks. It performs the fraud
check analysis of an URL and generates a warning message if URL is not secure. Parser 4
searches for other URLs whose pattern matches with the requested URL. It finds all details of the
other similar URLs and compares all the details (like year of domain registration, rating of the
domain, popularity of the domain etc.) with the requested URL details. It then displays all the
results in the preference on the browser screen before visiting the requested URL.
In implementation of parser 4 and 5, the Open Source Crawler “crawler4j” has been used. [17]
3.3. Re-visit Policy
In the proposed methodology, the parsers also use the re-visit policy when needed because web
has its dynamic nature. The re-visit policy can be easily understood using the freshness function
described in two ways viz. Freshness and Age. Freshness is used as binary measure which
indicates whether the local copy is accurate or not. The freshness of any page ‘p’ in the repository
at time t is defined as:

Age is a measure which indicates how outdated the local copy of page is. The age of a page ‘p’ in
the repository, at time t is defined as:
4. EXECUTION OF THE PROPOSED METHODOLOGY
Execution of proposed methodology depends on the sequence of knowledge bases and
corresponding parsers. Final result of the proposed methodology is not affected by the sequence
of the operations. Sequence affects only the space complexity and time complexity of the
methodology.
Execution of proposed methodology in divided into several steps which are described as follows
in the following sections
Figure 1. Flowchart of Step 1 of the proposed methodology
4.1 Historical Attack Detection
This step is composed with 2 operations which are occurred using ‘Knowledge Base I’ and
‘Knowledge Base T’.
Knowledge Base Initial (KBI) is used to detect the attacks which has the same pattern with the
previous detected attacks stored in it. Knowledge Base Trusted (KBT) is used to find the trusted
status of requested URL which was previously declared by the user. In Historical attack detection
the browser first tallies the URL with the KBI to check if its pattern matches that of any frequent
phishing attack stored in the knowledge base. If it is safe then it proceeds to match up with the
KBT. In this knowledge base it matches the URL against the trusted URLs stored by the user.
4.2 URL Pattern based Attack Detection
It is composed 2 operations which are related to ‘Knowledge Base A’ and ‘parser 1’. This Step 2
provides the security against those attacks which are purely URL-pattern based phishing as well
as SSL attacks. Knowledge Base A detects only those attacks which were detected previously by
the browser and were stored in its database. During the step 2, ‘parser 1’ scans the requested URL
and finds the occurrence of special characters (‘-‘,’.’ etc) and their repetition in the URL. It is
used to detect the pattern based phishing attacks. Working of step2 is represented in Figure 2.

4.3 URL Information Analysis
URL information can be very helpful in detection of the phishing attacks. This step is based on
the fact that the phishing URLs are newly registered and have lower rating and popularity over
the internet. Figure 3 represents the working of URL information analysis step. In this step
requested URL is analyzed with the Knowledge Base B and information of URL is analyzed
using the historical data of URL (if the URL was visited previously) and displays the results and
generates warning if URL is phishing attack based URL.
If the URL is not present in the history of Knowledge Base B then it goes to the parser 2 for the
information analysis. ‘Parser 2’ works to finds the information of URL as a web crawler (which is
described above) and performs the proper analysis after crawling for the details of the URL over
the internet.
4.4 Fraud URL Detection
This step is performed by the Knowledge Base C and parser 3.Knowledge Base C performs the
fraud check analysis of the requested URL (if it is available in the history of Knowledge base). It
displays the result and appropriate messages. If the URL is not visited previously then parser 3
performs the Fraud check analysis to provide security against phishing attacks (or other web
attacks) using some security algorithms. Figure 4 describes the fraud check analysis.

Figure 4. Flowchart of Fraud Check Analysis
5. IMPLEMENTATION AND RESULTS
We have implemented the proposed methodology in a virtual scenario, where we explored all the
visited URLs of browsers on different machines using the history feature. All the URLs have
been stored in a database for detect the phishing attacks and perform the analysis. We are
planning to implement this methodology with some new add-ons to install in present web
browsers (like other Firefox add-ons).
We have analyzed the URLs visited over the 5-months. In the initial stage of implementation,
security risks are more because of absence of data in the different knowledge base. The
implemented scenario provides 97.98 % security against phishing attacks and some hacking
attacks. We have not executed our proposed methodology for the duration of Dec, 2012 and Jan,
2013 but during Feb, 2013 to April, 2013, we have executed the above methodology.
The following table data represents the recorded activities of the Web URLs in the implemented
scenario and detected the phishing attacks and some hacking attacks.
Table 1: URL and some Web Attacks Analysis (2012-13)
Month Dec Jan Feb Mar Apr
No. of URLs visited 978 967 897 1023 1218
Phishing Attacks 21 18 17 22 24
Detected phishing attacks with the
browser
15 12 15 20 24
SSL Attacks 19 13 12 9 14
Detected SSL attacks with the browser 13 11 12 8 14
Execution Time (in minutes) 0 0 165 204 284
The above table shows the number of phishing attacks encountered and the execution time taken
by our methodology from December 2011 to April 2012. The execution time for the first two
months is actually zero as we have not implemented our methodology then. We have
implemented our methodology from February 2012 onwards.
Kindly note that the approximate time of execution per URL visit, for the first month comes out
to be 11 seconds. This increases to 12 seconds in the second month and to 14 seconds in the third
month. This gradual increase can be attributed to the fact that the knowledge base is increasing in
size hence the browser searches for more security attack then before

6. CONCLUSIONS
We have recorded the web URLs activities of with the usage of proposed methodology and
without usage of proposed methodology over 5 months. From the data, we have analyzed the
attacks and detected attacks over the time. Our system indicated that the 97.98% security against
phishing attacks as well as SSL-attacks over the browsing. Table 1 represents the recorded data
over the 5 months’ time period. Limitations of the proposed method are that due to various
parsing operations, its time complexity and space complexity is higher. So many times, it
increases the browsing time of web browser. Due to slower speed of browsing, generally web
users avoid this type of higher web security.
REFERENCES
[1] Ollmann G., The Phishing Guide Understanding & Preventing Phishing Attacks, NGS Software
Insight Security Research
[2] Aburrous ,Maher Ragheb, Alamgir, Hossain,Keshav Dahal, Thabatah, Fadi, "Modelling Intelligent
Phishing Detection System for E-banking Using Fuzzy Data Mining," cw, pp.265-272, 2009
International Conference on CyberWorlds, (2009)
[3] Abu-Nimeh, S.; Nair, S., "Bypassing Security Toolbars and Phishing Filters via DNS Poisoning,"
Global Telecommunications Conference, 2008. IEEE GLOBECOM 2008. IEEE , vol., no., pp.1-6,
(Nov. 30 2008-Dec. 4 2008)
[4] Yu, W.D.; Nargundkar, S.; Tiruthani, N., "A phishing vulnerability analysis of web based systems,"
Computers and Communications, 2008. ISCC 2008. IEEE Symposium on, vol., no., pp.326-331, 6-9
(July 2008)
[5] Alnajim, A. and Munro, M. 2009. An Anti-Phishing Approach that Uses Training Intervention for
Phishing Websites Detection. In Proceedings of the 2009 Sixth international Conference on
information Technology: New Generations (2009). ITNG. IEEE Computer Society, Washington, DC,
405-410. DOI= http://dx.doi.org/10.1109/ITNG.2009.109
[6] Chen ,Juan and Guo ,Chuanxiong, Online Detection and Prevention of Phishing Attacks, in Proc.
Chinacom 06
[7] Beginning PHP5, Apache, and MySQL Web Development by Elizabeth Naramore, Jason Gerner,
Yann Le Scouarnec, Jeremy Stolz, Michael K. Glass; ISBN: 9780764579660
[8] Sophos White Paper, Phishing and the threat to corporate networks,( 2005)
[9] PHP, AJAX, MySql and JavaScript Tutorials, http://www.w3schools.com/
[10] Prentice Hall - Deitel - Java How to Program, 4th Edition, Java_2_Complete_Reference_5E , Java -
How To Program, 6th Edition
[11] Ahn ,Luis von, Blum, Manuel, Hopper, Nicholas, and Langford ,John. CAPTCHA: Using Hard AI
Problems for Security. In Eurocrypt
[12] Gedam,Dhiraj Nilkanthrao,RSA Based Confidentiality And Integrity Enhancements in SCOSTA-CL,
A thesis report,Department of Computer Science and engineering,Indian Institute of Technology
,Kanpur, India, (July,2009)
[13] J. Ullman, Database and knowledge base systems, In Database and knowledge-base systems, volume
2, Computer Science Press, 1989
[14] Akerkar RA and Sajja Priti Srinivas: “Knowledge-based systems”, Jones & Bartlett Publishers,
Sudbury, MA, USA (2009)
[15] Delayed password disclosure Markus Jakobsson, Steven Myers, Palo Alto Research Center, 3333
Coyote Hill Road, Palo Alto, CA 94303, USA. School of Informatics, Indiana University,
Bloomington, IN, USA Journal: Int. J. of Applied Cryptography 2008 - Vol. 1, No.1 pp. 47 - 59
[16] Herzberg, A. and Gbara, A. (2004) Technical Report 2004-23, Protecting (even) Naïve Web Users,
or: Preventing Spoofing and Establishing Credentials of Web Sites, DIMACS, October 30, Available
at: http://dimacs.rutgers.edu/TechnicalReports/2004.html.
[17] The java project of the “crawler4j” can be downloaded from here:
http://www.code.google.com/p/crawler4j/

Authors
Gaurav Kumar Tak is an assistant professor in the School of Computer Engineering,
Lovely Professional University. His primary research areas of interest are Cyber-crime
and Security, Wireless Ad-hoc Network and Web Technologies. He has written several
research papers in these areas and continues to work for improving the security of web
applications and making the web safe to surf.
Gaurav Ojha is a student in the Department of Information Technology, Indian Institute
of Information Technology and Management Gwalior. His areas of interest are Web
technologies, Open source software and Internet Security.

Knowledge base compound approach against phishing attacks using some parsing techniques

More Related Content

What's hot (18)

Similar to Knowledge base compound approach against phishing attacks using some parsing techniques (20)

Recently uploaded (20)

Knowledge base compound approach against phishing attacks using some parsing techniques