Semantic approach utilizing data mining and case based reasoning for it support service

IJRET: International Journal of Research in Engineering and Technology eISSN: 2319-1163 | pISSN: 2321-7308
__________________________________________________________________________________________
Volume: 03 Issue: 03 | Mar-2014, Available @ http://www.ijret.org 358
SEMANTIC APPROACH UTILIZING DATA MINING AND CASE-BASED
REASONING FOR IT SUPPORT SERVICE
Niloofer Shanavas1
, Shimmi Asokan2
1
M.Tech. Student, Department of Computer Science & Engineering, Rajagiri School of Engineering & Technology, Kochi,
Kerala, India
2
Assistant Professor, Department of Computer Science & Engineering, Rajagiri School of Engineering & Technology,
Kochi, Kerala, India
Abstract
Information Technology (IT) plays a very important role in all organizations. IT executives are constantly faced with problems that
are difficult to tackle. Failure in IT service can interrupt the functioning of an organization. Case-Based Reasoning (CBR) is a
problem solving methodology where experience in the form of past cases can be used to solve problems, thereby assisting the
automation of problem solving and experience management. Furthermore, the performance, quality and efficiency of CBR systems
can be enhanced through data mining. In order to support the IT team for faster and efficient problem resolution, a case-based
reasoning approach integrated with data mining techniques could be utilized. In this paper, the study done on various CBR systems
and data mining techniques for problem and experience management is explained. A system is proposed for IT experience and
problem management with semantic retrieval in order to increase the efficiency and quality of the IT support service.
Keywords: Case-based reasoning, Data Mining, Experience management, IT problem management, IT support.
-----------------------------------------------------------------------***----------------------------------------------------------------------
1. INTRODUCTION
Today, Information Technology is vital for most of the
organizations and an important element for their functioning.
Every organization has to invest in Information Technology to
compete and succeed. Gathering and sharing IT knowledge
can increase the organizational intelligence, allowing
organizations to compete much better [1]. IT is strongly
embedded in the business and has become an integral part of
business management. Decisions on IT matters have brought
about far reaching consequences for the business. IT
executives are burdened with questions that are not easy to
tackle. The IT team can resolve issues quickly and effectively
if they have the information that can support them in their
problem resolution. Collection of relevant data, analyzing it
and proper interpretation is needed for obtaining this
information [2].
The increasing significance of Information Technology
resources and associated services that supports the
organization´s business units have made Information
Technology Service Management (ITSM) of vital importance
for an organization [3]. Information Technology Infrastructure
Library (ITIL), a methodology used in large organizations for
IT management, is a widely accepted approach for IT Service
Management across the world. ITIL provides a practical, no
nonsense framework for identifying, planning, delivering and
supporting IT services in the business [4]. ITIL framework
emphasizes on Problem Management to reduce the number
and severity of incidents, thereby reducing potential problems
to the business/organization. Problem management consists of
reactive and proactive aspects. Solving a problem when one or
more incidents arise is the reactive problem management [5].
Identifying and solving potential causes of error before any
incidents occur is the proactive problem management. Reusing
IT knowledge and problem resolution history is of high value
to reach these problem management goals. This helps us to
detect the root causes of IT problems in a more effective way
and help bring down the associated costs. In [7], case-based
reasoning is defined as a reasoning model that assimilates
problem solving, understanding, and learning, and integrates
all of them with memory processes. CBR functions with the
help of case-based knowledge. In [7], a case is defined as a
contextualized piece of knowledge called experience that is
required to attain the goals of the system. Comparisons
between the new and old situations are used in a case-based
system. This method of case-based reasoning depends on the
assumption that “similar problems have similar solutions”.
This assumption that is referred to as the CBR hypothesis is
the managing principle essential for most CBR systems [8].
Since case-based reasoning resolves new issues by using
resolutions used to solve past issues, it is extensively used to
develop support systems for IT management. Since it is
possible that past issues can arise again in a similar fashion,
CBR systems are very useful for problem management. Thus
CBR automates the process of describing problems and
defining solutions to these problems [9].

__________________________________________________________________________________________
The CBR architecture consists of four prime components as
shown in Fig. 1 [7]. In CBR system, similar previous cases are
retrieved to solve a new problem. The retrieved cases suggest
a solution for the problem. If necessary, the suggested solution
can be revised for solving the new problem; this then becomes
the confirmed solution. The new solution is retained as part of
a new case and is called the learned case. This is the core
process of the CBR cycle.
Fig 1 CBR cycle
With the help of data mining technology for feature selection
and case retrieval, the quality of CBR systems can be
improved [10]. A huge case consisting of a large number of
features can be reduced to a feature vector with only a few
principal features through feature selection. In a case base, the
cases can be clustered into sub-case bases at different levels
that make case management simple, case retrieval more
efficient and accurate, thereby increasing the efficiency of
CBR systems. Data Mining is an efficient and significant tool
for knowledge discovery from a large database. New
approaches to problem solving are obtained with the help of
data mining for decision support by discovering patterns and
relationships hidden in data.
2. IT PROBLEM MANAGEMENT
IT Problem management is one of the components in the ITIL
support area. Problem management helps in reducing the
number of incidents and potential problems in an organization
[5]. Problem management minimizes problems and incidents
due to errors within the IT infrastructure. Problems are
assigned a priority level with the highest priority given to a
problem that can interrupt critical IT services. Prioritization
helps the IT support team to identify the impact and urgency
associated with problems. If problems are not detected and
solved, it can have a serious impact on the availability of IT
services. An important task of problem management is to
document the information about the incident in order to
support all problem management activities. Problem
management has many advantages. It improves IT service
quality, reduces the number of incidents, obtains permanent
solution, improves organizational learning and increases fix
rate [5]. Reliable IT services are very necessary for an
organization. Incident reductions minimize the disruptions in
the organization. Permanent solution for problems gradually
reduces the number of problems. Problem management
depends on learning from previous solved cases. Service desk
can easily solve problems that are solved earlier and
documented. An effective system that aids the IT executive in
solving problems is essential for the success of problem
management. In [6], a solution is proposed to find the root
cause of problems from previous diagnoses of IT failures.
Through interaction with IT operators, it finds the root cause
of problems. Hence, importance is given to the reuse of
knowledge and IT operator’s experience. Earlier solutions
were static, making it impossible to reuse knowledge to solve
similar failures. Failures have to be corrected without any
delay to maintain high quality service. The steps in problem
resolution are observing service alerts, raising trouble tickets,
assigning tickets, identifying root cause, planning resolution,
implementing resolution and notification/clean up [15]. The
major troubleshooting effort is spent on identifying the root
cause and planning resolution. The troubleshooting effort
needs to be reduced to save the time and number of resources
required for problem resolution.
3. EXPERIENCE MANAGEMENT
In this busy and competitive world, one area that has gained
prominence is experience management. Experience is required
for complex problem solving. Experiences are shared to obtain
the experiences of different people and avoid the kind of
mistakes that other people have done. The problem need not
be solved from scratch by reusing experience. Experience
management is a particular kind of knowledge management
[12]. A problem solver obtains specific valuable knowledge
called experience while problem solving. The specific
activities of experience management are to collect, model,
store, reuse, evaluate and maintain experience. The experience
base is a collection of experience items. In case-based
reasoning, experience item is known as the case and case base
is the other term for experience base. Approaches towards
experience management include Experience Factory and
Lessons Learned System. An approach called Experience
Factory, particularly tailored to software engineering,
captures, documents, stores and disseminates experience [11].
These experiences are stored in an experience base that forms
organizational memory. Hence, the experience factory is an
organizational framework that supports reuse and learning.
Another system that implements knowledge management
approach to collect, store, disseminate and reuse experiential
knowledge is the lessons learned system [23]. CBR
methodology that obtains solution of a new problem from

__________________________________________________________________________________________
similar problems is applied as a foundation for organizational
systems that manage experience [17].
4. CASE-BASED REASONING (CBR) SYSTEMS
CBR systems make use of the specific knowledge of past
experiences (cases). A similar past case is found for a new
problem and reused to obtain the solution for the new
problem. In [13], it is mentioned that CBR is also an approach
to incremental learning since a new experience obtained by
solving a problem is made available for future problems. In
[14], a system called DUMBO is presented that integrates
case-based reasoning methodology in traditional trouble ticket
system architecture that extends its functionalities by
diagnosing new problems using previous network faults.
DUMBO performs continuous learning, i.e. the knowledge
base is gradually increased by the integration of new solved
cases. It uses production rules in the reasoning process. Other
systems that use cased-based reasoning in association with
trouble ticket systems are MASTER and CRITTER [14]. A
positive feature of CRITTER is that its adaptation strategy is
automated and that of MASTER is its friendly user interface.
The advantages of DUMBO compared to these systems are its
simple structure, adds new relevant features in the learning
and allows users interaction [14]. The disadvantage of
DUMBO system is that the production rules for case matching
are manually generated.
In [15], the information retrieval component of system is
described that retrieves relevant articles from a collection of
previously solved problems and their associated solutions.
This system assists the system administrators in problem
resolution by identifying similar problems and retrieving the
steps taken to resolve those problems. A term vector is used to
represent an article. Taxonomy is used to define the document
representation space and it gives a source of keywords for the
initial case description.
In [16], a web-based dynamic knowledge base system called
NetPal is described. NetPal supports network administrators to
troubleshoot tasks, recall and store experience, and identify
new failure case and hence reduces time and resources
required. It supports experience management and assists
system administrators to recall by suggesting past experience
cases. The system represents various domains like knowledge
management, information retrieval, machine learning and
network management. The vector space model is used to
represent the cases for information retrieval and cosine angle
distance is used to measure the similarity between vectors.
Log Data Acquisition Engine collects and preprocesses log
data from multiple services that are aggregated in a central
database and indexed for retrieval. One drawback is that
experience cannot be automatically collected for the case
database.
5. DATA MINING FOR IT EXPERIENCE
MANAGEMENT SYSTEMS
Data mining, also known as knowledge discovery in databases
(KDD), is an emerging field. It helps to analyze, understand
and visualize the vast amounts of stored data gathered from
several business applications and help organizations to make
better decision to stay competitive in the industry [21]. The
introduction of data mining to information systems improves
the quality of knowledge discovery process and decision
process [22]. The techniques of data mining like clustering,
classification, association rules and generalization could be
used to optimize and improve Experience Management
Systems [11]. Clustering is used to group similar experiences.
Classification is used for the categorization of experiences into
classes. Association rules are used to find relationships
between data in a large database and generalization is used for
data summarization. These data mining techniques are used in
analyzing, generalizing, processing the experiences and
building the experience base [11]. Frequently asked questions
(FAQ) pages assist IT executives in responding to inquiries. In
[18], a system is proposed that help a service company’s
operators respond to inquiries and construct FAQ. The system
has four components: FAQ filtering, FAQ matching, replying
support and FAQ building. The FAQ filtering part performs a
comparison of inquiries with the questions in the FAQ to
determine whether they are related. The FAQ matching part
automatically replies with a question and answer set in FAQ if
the inquiry is related to a question in the FAQ. The inquiries
that are not related to any question in the FAQ are forwarded
to the operators. The replying support component displays a
reply form to operators based on inquiry and replying records
from the past. In the FAQ building part, operators are
displayed a FAQ that are constructed from past inquiries.
Clustering technique is used in the FAQ building part to group
similar inquiries and answer sets so that it becomes easier for
the operators to analyze the records that can save their time. In
[19], it is mentioned that there could be large amount of cost
saving if the IT support team reduces the problem
troubleshooting time. The proposed solution in [19] is an
information retrieval approach combined with data mining
techniques to extract information automatically from public
resources such as FAQs and forums to generate a knowledge
base for dynamic IT management support. This is especially
beneficial for small to medium sized organizations that do not
have many experienced IT staff. Crawlers are employed to
collect experience data (problem/solution pairs) from publicly
available resources such as FAQs, forums and web sites that
are useful for the IT management team. The data is
preprocessed by punctuation removal, stop word removal and
stemming; each data instance is represented by a term vector
using its term frequency/inverse document frequency
(TF/IDF). Then, clustering techniques are used to group
similar problem/solution pairs for quick retrieval. The
clustering algorithm used is an improved version of that used
in [18] as it reduces its computational cost. In [20], the

__________________________________________________________________________________________
proposed system in [19] is improved by using the clustering
algorithm with Multi objective genetic algorithm (MOGA), an
optimization technique that automates the process of choosing
clustering parameters.
6. PROPOSED SYSTEM
The studies conducted on data mining and case-based
reasoning for IT support service, reveals that keyword-based
retrieval has many limitations compared to concept-based
retrieval/semantic retrieval. Synonymy (multiple words having
similar meaning) and polysemy (words having more than one
meaning) are the main causes for inaccurate retrieval in
keyword-based systems since this results in mismatches
between the vocabulary used by users and that used in the
documents. An effective approach towards concept-based
retrieval is to use Latent Semantic Indexing (LSI) / Latent
Semantic Analysis (LSA), a soft clustering algorithm and a
dimensionality reduction technique. Latent Semantic Indexing
[24] is a statistical information retrieval method that retrieves
text based on matching of concepts and not that of keywords.
We propose a system with semantic retrieval as shown in Fig.
2 for IT problem and experience management that increases
the efficiency and quality of the IT support service.
Fig 2: IT Experience and Problem Management System
Design
The web crawler extracts IT experience data or solved IT
problems from public resources. The extracted information is
preprocessed to obtain the dataset containing problem/solution
pairs. Latent semantic analysis is done on the documents in
order to transform it into lower dimensional latent space. This
brings out the hidden structure of these documents and
represents the documents in a semantic way. The IT team can
semantically retrieve documents (problem/solution pairs)
corresponding to their problem. They can also add new
problem/solution pairs to the knowledge base which results in
a dynamically incrementing IT experience knowledge base.
The proposed system guarantees accurate and efficient
retrieval of similar problem/solution pairs from the IT
experience knowledge base.
7. CONCLUSIONS & FUTURE WORK
Information Technology support is an area that requires vast
amount of knowledge. Trouble shooting time could be reduced
if the IT experience gained could be accumulated in an
experience base. Hence, IT experience management is
valuable. Case-based reasoning approach which is guided by
the principle that similar problems have similar solutions is
widely used to develop support systems for IT management
that solves new problems by using or adapting solutions used
to solve old problems. This paper explains how CBR systems
integrated with data mining techniques can automate IT
experience and problem management and proposes an
efficient system for IT support service. Focusing more on
approaches for semantic processing can continue research in
this area.
REFERENCES
[1] Paul Klint, Chris Verhoef, “Enabling the creation of
knowledge about software assets”, Data and
Knowledge Engineering e41 (23), pp. 141-158, 2002.
[2] L.M. Kwiatkowski, C. Verhoef, “Recovering
management information from source code”, Science
of Computer Programming, Volume 78, Issue 9, pp.
1368-1406, 2013.
[3] Jacques P. Sauve, Rodrigo A. Santos, Rodrigo R.
Almeida, J. Antao B. Moura, “On the Risk Exposure
and Priority Determination of Changes in IT Service
Management”, 18th IFIP/IEEE International Workshop
on Distributed Systems: Operations and Management
(DSOM 2007), pp. 147-158, 2007.
[4] http://www.itil-officialsite.com/
[5] http://www.ucisa.ac.uk
[6] Ricardo L. dos Santos, Juliano A. Wickboldt, Roben C.
Lunardi, Bruno L. Dalmazo, Lisandro Z. Granville,
Luciano P. Gaspary, Claudio Bartolini, Marianne
Hickeyt, “A solution for identifying the root cause of
problems in IT change management”, Integrated
Network Management (IM), 2011 IFIP/IEEE
International Symposium, pp. 586-593, 2011.
[7] Shiu S., Pal S. K., “Foundations of soft case-based
reasoning”, John Wiley & Sons, 2004.
[8] Hullermeier E., “Toward a probabilistic formalization
of case-based Inference”, Proceedings of the 16th
international joint conference on artificial intelligence
(IJCAI99), Stockholm, Sweden, pp. 248-253, 1999.

__________________________________________________________________________________________
[9] T. Olsson, P. Funk, “Case-based reasoning combined
with statistics for diagnostics and prognosis”, 25th
International Congress on Condition Monitoring and
Diagnostic Engineering, Journal of Physics:
Conference Series, 2012.
[10] Yuan Guo, Jie Hu, Yinghong Peng, “Research on CBR
system based on data mining”, Applied Soft
Computing Volume 11, Issue 8, pp. 5006-5014, 2011.
[11] Klaus-Dieter Althoff, Bjorn Decker, Susanne Hartkopf,
Andreas Jedlitschka, Markus Nick, Jorg Rech,
“Experience Management: The Fraunhofer IESE
Experience Factory”, Proc. Industrial Conference Data
Mining, 2001.
[12] Bergmann Ralph, “Experience Management:
Foundations, Development Methodology, and Internet-
Based Applications”, Springer, 2002.
[13] Aamodt, E. Plaza, “Case-Based Reasoning:
Foundational Issues, Methodological Variations and
System Approaches”, AI Communications, IOS Press,
Vol. 7: 1, pp. 39-59, 1994.
[14] Melchiors, L. M. R. Tarouco, “Troubleshooting
network faults using past experience”, Network
Operations and Management Symposium, pp. 549-562,
2000
[15] George A., Makanju A., Zincir-Heywoo A.N., Milios
E.E., “Information Retrieval in Network
Administration”, Communication Networks and
Services Research Conference, pp. 561-568, May 2008.
[16] Ashley George, Adetokunbo Makanju, Evangelos
Milios, Nur Zincir-Heywood, Markus Latzel, Sotirios
Stergiopoulos, “NetPal: a dynamic network
administration knowledge base”, CASCON ’08:
Proceedings of the 2008 conference of the center for
advanced studies on collaborative research, 2008.
[17] Klaus-Dieter Althoff, Rosina O. Weber, “Knowledge
management in case-based reasoning”, The Knowledge
Engineering Review, Vol. 20:3, Cambridge University
Press, pp. 305-310, 2006.
[18] Iwai K., Iida K., Akiyoshi M., Komoda N. “A help
desk support system with filtering and reusing e-
mails”, Industrial Informatics (INDIN), 2010 8th IEEE
International Conference, pp. 321-325, 2010.
[19] Can Bozdogan, Nur Zincir-Heywood, “Data Mining for
Supporting IT Management”, Network Operations and
Management Symposium (NOMS), IEEE, pp. 1378-
1385, 2012.
[20] Bozdogan C., Zincir-Heywood A.N., Gokcen Y.,
“Automatic optimization for a clustering based
approach to support IT management”, Integrated
Network Management (IM 2013), 2013 IFIP/IEEE
International Symposium, pp. 1233-1236, 2013.
[21] S.C. Hui, G. Jha, “Data mining for customer service
support”, Information & Management, Volume 38,
Issue 1, pp. 113, October 2000.
[22] Rok Rupnik, Matjaz Kukar, “Data Mining and
Decision Support: An Integrative Approach”, Decision
Support Systems, Book edited by Chiang S. Jao, 2010.
[23] R. Weber, D.W Aha, I. Becerra-Fernandez, “Intelligent
lessons learned systems”, Expert Systems with
Applications, Volume 20, Issue 1, pp.17-34, 2001.
[24] Scott Deerwester, Susan T. Dumais, George W.
Furnas, Thomas K. Landauer, Richard Harshman,
“Indexing by latent semantic analysis”, Journal of the
American Society for Information Science, pp. 391–
407, 1990.

Semantic approach utilizing data mining and case based reasoning for it support service

More Related Content

What's hot (16)

Viewers also liked (18)

Similar to Semantic approach utilizing data mining and case based reasoning for it support service (20)

More from eSAT Journals (20)

Recently uploaded (20)

Semantic approach utilizing data mining and case based reasoning for it support service