Description
Business Intelligence Modeling A Case Study of Disaster Management Organization in Pakistan
Business Intelligence Modeling: A Case Study of Disaster Management
Organization in Pakistan
Sohail Asghar
Department of Computer Science
Mohammad Ali Jinnah University,
Islamabad, Pakistan
e-mail:
[email protected]
Simon Fong
Department of Computer and
Information Science
University of Macau
Macau SAR
e-mail:
[email protected]
Touqeer Hussain
Shaheed Zulfikar Ali Bhutto
Institute of Science and
Technology,
Islamabad, Pakistan
e-mail:
[email protected]
Abstract—Acquiring useful Business Intelligence (BI) for high-
qualitative decision making is a challenging task in today’s
dynamic business environment. Powerful tools are available
such as ETL, data-warehouse, OLAP, data-mining and
visualization. However, companies are lack of standardization
of procedures in cultivating BI for improving the performance
of the organization at all levels. In this paper, we developed a
model that link dimensions of BI and processes together, for
providing a good decision support, for disaster management
organization in Pakistan as a case study. The model is
implemented and validated using Oracle BI tools and
techniques. The significance of this research work is the BI
model proposed which provides exploratory abilities on the
data, and linkages among BI processes from the conceptual BI
dimensions.
Keywords-Business Intelligence dimension, Data Mining,
OLAP, Data Warehouse, and Business Functionaity
I. INTRODUCTION
BI is recognized as an increasingly important support for
business decision making [1] in today’s emerging business
environment, where a huge amount of data is growing fast
and scattered around. Most organizations nowadays face the
challenges of processing and analyzing enormous amount of
data [4] for real-time BI. To tackle such problems, data are
progressively transformed into useful information and
knowledge management techniques are implemented to
manage the information and to support decision making [5].
In assimilation to above BI also supports different
enterprises by utilizing its tools and techniques [6] in order to
support them in carrying out different processes and handle
the processes according to different tasks. In regard of this
BI is identified as an amalgamation of reporting, data mining
and online analytical processing applications. BI supposed to
provide access to data that has been cleaned and integrated
so that they can be analyzed, manipulated, transformed, and
combined to discover correlations, trends, and patterns that
offer new insights, aid in decision-making. However,
enterprises adopt different practices for BI by using a mix of
ETL, data warehouse, OLAP, data mining and decision
models [1][4][8]. Data warehouse is the core of BI system
which stores aggregated and historical data [1]. It is loaded
from many operational data sources i.e. MIS, CRM, ERP and
other legacy systems [1][4][6], by some automated
procedure called ETL (Extract, Transform and Load) [7]. On
top of that, OLAP (Online Analytical Processing) provides
analytical views of the stored data in the warehouse in the
form of star or snowflake schema [1][3][4]. On the other
hand, Data Mining helps to find hidden patterns from the
data through different algorithms (association rule mining,
decision tree, and clustering etc).
With fast growth of business data in both volume and
variety, amalgamation of data from different sources by
different autonomous processes into the corresponding BI
dimensions is a challenge. To cope with that, we have
proposed a model which essentially provides an exploration
and linkage among the BI processes and its dimensions for
good decision making. In this paper we describe a detailed
validation and implementation of the model through a real-
life case study. Some key tools and techniques used include
Oracle 11g, Oracle analytical work manager (for OLAP) of
data warehousing for its multidimensional cube generation.
For data representation in cubes as a user interface, some
operations have been carried out like slicing, dicing, rollup,
drill down and that of pivoting. The proposed model is
composed of two main parts: BI dimension and BI process.
BI dimension embraces the management concepts of
knowledge, functionality, technology, business and
organization; whereas BI process includes the technical
activities of operational data sources, ETL, data warehouse,
OLAPS, data mining and visualization tools.
Consequently, we have linked the BI dimension with the
BI process in our model, which is essential during the life
cycle of BI system development. As a result of the
integration it is possible for us to implement efficient
architecture by quick prototyping. The developed prototype
architecture is then tested with real life data. The
experimental results are validated and we highlight the
benefits attained.
The remaining portions of this paper are organized as
follows: Section 2 presents a literature review of current
research efforts by various researchers in the area of BI
models and methodology. We show our proposed
architecture of BI system in Section 3. Section 4 discusses
the validation and evaluation of experimental results. Finally,
Section 5 provides a conclusion and possible future research.
II. LITERATURE REVIEW
In 2003, Olszak and Ziemba [1] suggested an integrated
approach to build and implement business intelligence
solution. This integrated approach composes of four basic
dimensions i.e. business, function, technology and
organization. According to the authors an adequate approach
is required to design, implement and use the BI system,
keeping in mind the four dimensions where BI systems are
being initiated. The same approach has been suggested by
the [2] and [7]. Our approach is similar with respect to their
proposed BI dimension. But we are considering also the
knowledge as an additional dimension. We explored these
dimensions and linked them with BI creation,
implementation and using the process as shown in Fig. 1. We
implemented this approach in practical domain and
developed a BI application whereas [1] only suggested the
implementation approach.
Later in 2004, Simmer [2] proposed a stakeholder model
of business intelligence which was closely related to the
model presented in [1]. According to the author for the
purpose of expansion in the organization performance, this
model is potentially useful for design, diagnosis, and
enhancement of BI because it integrates stakeholder system
with technical and human knowledge systems. The
stakeholder model of BI is mainly divided into two parts:
external environment and internal environment. Whereas the
external environment covers the institutional, extra-
institutional and industry environment and internal consists
of business practices, knowledge base, knowledge source,
value creation, and strategic attributes. However, the
connection of stakeholder model with respect to our
proposed model is similar based upon the BI dimension. The
external factors of stakeholder model which we have covered
into our organizational dimension and other dimension are
covers into internal factors. An important feature of our
model is that it is quite generic because we have also focused
on the BI process and extensively explored it which is the
limitation of the model presented in [1] and [2].
Xu et al [3] presented BI applications infrastructure for
the life insurance industry. They have demonstrated that how
an innovative BI infrastructure and application could
effectively address major business challenges. Here arises a
question of how this will help us to achieve operation
excellence and business impact. The major parts of the
infrastructure are meta-DBs, which are operational data
sources. The second part is a BI application that includes
data mining, OLAP, data queries and data warehouses. The
last part is BI users including end-users of the application,
based on their roles/responsibilities. The presented work is
limited in relation to our proposed model that it cannot
identify the transformation of data from operational data
sources to data warehouse, which is an essential part of the
BI system. Our proposed model however highlights on the
transformation procedure from operation data sources to data
warehouse, which is illustrated in Fig. 1.
In 2006 Zeng et al. [4] claimed that successful
application of BI in an enterprise should be measured
through two reasons. The first is correct, valid, integrated
and in-time data, and the other is the means which
transforms the data into decisions. Both tasks have certain
difficulties. Therefore the authors presented a BI technical
framework which consists of an operational application tier
(legacy data sources), a data acquisition tier (ETL), a data
warehouse tier, BI suite tiers and corporate performance
management tiers. The model is quite similar to our
proposed model except that our model has details in
relationship of BI dimensions and BI processes which
mutually facilitate decision making.
In the fall of 2006, Ou and Peng [5] introduced the
concepts of business process management to the current
business intelligence with the ability of process-driven
decision making. These processes are stored in process
model base and are flexible and reusable. By using case
based reasoning technology, matching processes can be
retrieved and delivered to the decision makers when they
encounter a new problem. The authors tried to address the
situation when the existing data sources could not fulfill the
problem statement of new problem. In this situation the
assistance of knowledge implementation and business
process reuse can attain the decision making efficiency of BI
system. To deal with these issues the author proposed a
knowledge base business intelligence system (KBBIS). With
respect to our proposed model the knowledge dimension also
deals with such types of issues when the existing data
sources are insufficient to fulfill the problem statement.
Delibasic in 2007 [6] presented an abstract level model of
business intelligence system. It proposes to reach for
decision making through knowledge in two ways i.e. with
experts and computational supports. At experts side a
decision maker has support from the GDSS (group decision
supports system) with experts’ collaboration. The main parts
of the model are databases for the computation support,
various knowledge discovery tools (case based reasoning
system, data mining etc.) and visualization techniques. Our
model is supporting these features which are closely related
to the model presented in [6]. It also covers other aspects of
BI system which are not discussed by the author. Our model
covers a comprehensive range of dimensions such as
business, technology, functionality and organization, which
are very important in designing the BI systems.
Olszak and Ziemba in 2007 [7] introduced a
methodology of BI system creation and implementation for
organizations. This approach is focused on objectives and
functional areas of BI in organizations. According to the
authors, the BI has two major stages that are of interactive
nature, (1) BI creation and (2) BI consumption. The creation
stage involves a series of tools and technologies, which
include ETL, data warehouse, OLAP, data mining and
different presentation tools. Consumption stage involves the
fundamental changes in a particular enterprise. The authors
focused on the issue that organizations require some cultural
background to go along with information system and
information technology, when building and implementing a
BI system. This suggested methodology of building and
implementing BI system also need sound business practices
set by the enterprise. Our proposed model does not only
focus on the above listed issue, it actually provides a real
time application in the disaster management domain and
provides a better decision-making environment.
Zhong et al. in 2008 [8] applied BI into a social security
system. It provides business analysis to the decision makers.
Authors proposed a three-dimensional framework of BI
which is based on a three-level screening funnel. It refers to
the dimension of the user roles, the dimension of instruments
and the dimension of contents. In terms of three-level
screening funnel it advocates that it is necessary to screen by
industry positioning, positioning of value chains and
positioning of the stage for expanding users. This idea leads
to the best match of BI applied, namely carrying out concrete
operations. The role dimension is made up of three kind of
personnel namely, business personnel, technical personnel,
and business technical personnel. The content dimension
reflects application, inclusive of deductive application,
inductive application and the application for controlling and
developing the instrument dimension.
III. PROPOSED BUSINESS INTELLIGENCE MODEL
The proposed architecture is divided into two main parts
i.e. BI dimension and BI process. BI dimension includes the
concept of knowledge, functionality, technology, business
and organization whereas the BI process includes the
activities of operational data sources, ETL, data warehouse,
OLAPS, data mining and visualization tools. As a holistic
approach, we linked the BI dimension with the BI process as
a whole model which is essential during the complete life
cycle of BI system development.
A. BI Knowledge aspect
According to [1] knowledge is asserted for enterprise
which is applied in all key business processes. It also
constitutes a prerequisite for the development of new
products and technologies, volume of scales, reaching new
customers and maintaining relations with existing customers.
The originating source of knowledge in an enterprise
includes: their information systems, internal documentation,
media press, reports, domestic and foreign statistics, internet,
corporate databases, customers, suppliers, business partners,
and their employees [1][2]. However, knowledge can be
classified into following categories i.e. procedural,
descriptive, semantic, episodic, explicit and tacit knowledge
[2][6]. As a result, it is very important for an enterprise to
analyze the source of knowledge, data sources, knowledge
type and solution for the problem when existing data sources
are unable fulfill the requirement for the BI system.
B. BI Functional aspect
The second important dimension which is linked with the
BI process is the functional aspect. According to [1] BI
standard should be analyzed taking into consideration of all
the benefits which are likely to be generated in an enterprise.
Following are the functional aspect that is considered during
BI process [1]:
• Strategic planning included:
o Different variants in the development of an
organization
o Relation of an enterprise strategy, mission, goals
and tasks
o Identification of bottlenecks to be tackled
o Providing information on the enterprise
environment & market trends
• Improving relation with customers
• Analyzing profitability of products
• Analyzing the internal processes and operational
efficacy
• Controlling and managing accounting
Figure 1. Business Intelligence model.
C. BI Technology aspect
The third aspect of BI system is technology. According
to [1][4] different world-wide leading BI vendors provide
end-to-end advanced enterprise platforms; BI solution
primarily involves method of knowledge creation, source of
knowledge and information technology tools. However,
integration between world-class databases management,
analysis server, enterprise sources and flexible, easy-to-use
front-end applications make the power of BI accessible to
users at all levels of the organization and realize benefits in
some key access.
More and More enterprise solution and platform for BI
have been developed. Some commercial solution include,
just to name a few, Oracle databases, IBM DB2, Microsoft
SQL Server [1][4][7], with the BI tools Microsoft analysis
server, Oracle Analytical Work Manager (AWM), Oracle
Warehouse Builder [1], Oracle BIEE (Business Intelligence
Enterprise Edition), NCR teradata warehouse, Hyperion,
SAS, Cognos, Business Objects [7], Open source project
Pentaho, and MS Miner [4].
However, selecting appropriate BI tools from the above
list vender is not an easy task. The authors in [7] suggest a
set of criteria that should be taken into consideration during
the selection. The criteria are based on the following factors
that are being analyzed: functionality, complexity of
solution, compatibility, and organizational functional needs.
However, it is also important that these tools meet the
organization expectations in a foreseeable future.
D. BI Business aspect
The fourth important aspect of BI system is business.
According to [2] the increasing use of business practices has
given rise to spiraling demands for information. BI is able to
codify and share these demands through multiple internal
and external environments. In turn technical development
within BI has enabled the increased diffusion of these
business practices.
E. BI Organization aspect
The last but not the least important aspect of BI system is
organization. It is imperative to understand the business,
goals of an enterprise before implementing the BI solution
[1] because the present and future demands for knowledge in
an enterprise are based on these goals. Next tasks are to
analyze and design key processes, positions, responsibilities,
and stream knowledge flows.
IV. VALIDATION AND EVALUATION
This section is presented in five phases: in the first phase
we explain the implementation model details and elaborate
the different parts of the model; in the second phase we
enlighten the data warehouse schema, its development in
Oracle Warehouse Builder (OWB) and also the ETL process;
in the third phase we clarify the implementation details of a
case study with Oracle BIEE tool; in the fourth phase we
explain the ad-hoc queries implementation details with
Oracle Discover and Discover Plus (that are embedded in
Oracle Application Server 10g); we explain the integration of
Oracle BI solution with MS-Office which is very useful for
different analysis purposes in the last phase.
The implementation model is depicted in Fig. 2 and it is
divided into four main parts: the first part is operational data
sources which include the three main data sources physical,
financial and training. We had applied ETL process on these
data sources and the lode data into data warehouse shown in
Fig. 3. We mapped them on a logical schema called data
warehouse schema shown in Fig. 4. The schema consists of
three main facts tables and two dimensions which are TIME
and LOCATION respectively.
On the third step of implementation model we had linked
Oracle BIEE, Oracle Discover plus & Desktop, and MS-
Office with warehouse for data presentation. This
presentation comes with web browser and desktop utilities.
Fig. 5 gives a preview of business model development in
BIEE administration tools. This model contains three main
layers, i.e. physical, business model & mapping, and
presentation. Physical layer consists of actual data sources
imported from a data warehouse which is consisting of
dimension and facts tables. Based on this physical data
source we built a business model & mapping layer where we
could logically define a hierarchy of LOCATION and TIME
dimensions that are depicted in Fig. 6. When the hierarchy is
defined, it can be published in the presentation layer.
Figure 2. Business Intelligence Implementation Model.
Figure 3. Data Transformation Services (SQL Loader) wizard view.
Figure 4. Schema Development in Oracle Data Warehouse 11g (OWB)
Screen view.
After completing these steps we established an OC4J
server and named it BI Presentation Services which is a web
based interface shown in Fig. 6. In this interface we
compiled the different answers and save them into
appropriate folders, which in turn can be used as useful
queries presented on the dashboard as shown in Fig. 7
Figure 5. Business Model Developments in BIEE Administration Tool.
Figure 6. BIEE Answer Development wizard.
Figure 7. Real Time Activities Monitoring Dashboard.
Figure 8. Business Model Developments in BIEE Administration Tool.
Figure 9. Integration with MS-Excel 2003 as BI viewer.
The interactive dashboard is very useful for analysis
purposes. Users can roll-up and drill-down any answer
presented on the dashboard and move into the next level of
information which is in deeper details by a given domain. In
our case study, our users can navigate information up to four
levels: Province District Tehsil Union Council.
The ad-hoc query presented in Fig. 8 is built in Oracle
Discover Plus. User can build different cross-tabs and
multiple graphs from which BI reports can be printed.
Fig. 9 and Fig. 10 demostrate the integration service of
Oracle BI tools with MS-Excel and MS-Power Point
respectively. It is effective for the users that they can easily
call a query from the Oracle dashboard and display them in
BI viewers with the data in the MS-Excel formats.
Figure 10. Integration with MS-Power Point 2003 as BI.
V. CONCLUSION AND FUTURE WORK
In this project we linked BI dimension with BI process
which is a primary source of BI solution creation in any
organization. We have devised a architecture for the BI
system and made it more useful with this mapping of BI
process and dimension. Furthermore, the proposed
architecture along with its components is discussed,
highlighting their limitation and advantage. A prototype has
been developed and experiments are performed on a case
study of disaster management organization in Pakistan.
It is observed that our architecture is possible to
implement and manage. We experimented with several sets
of data, the results show that they are quick and efficient
decision-making can performed over the BI system.
Currently we are trying to explore the relation of
enhanced OLAP, different data mining techniques, real time
data warehousing and DSS. We are now in exploratory phase
of how these technologies contribute the BI system for
increasing its performance and visualization. The ultimate
goal is to develop a better business decision-making
environment.
REFERENCES
[1] C. M. Olszak and E. Ziemba, "Business Intelligence as a key to
Management of an Enterprise," Proceedings of Information Science
and IT Education Conference, pp. 855-863, June 2003
[2] Claire A. Simmers,"A Stakeholder Model of Business Intelligence,"
Proceedings of the 37th Hawaii International Conference on System
Sciences, pp. 1-9, 2004
[3] Z. Xu, M. Zhang and X. Jiang, "Business Intelligence - A Case Study
in Life Insurance Industry," Proceeding of the International
Conference on e-Business Engineering (ICEBE), pp. 129, 2005
[4] Li Zeng, et al, "Techniques, Process and Enterprise Solutions of
Business Intelligence," IEEE Conference on System, Man, and
Cybernetics, pp. 4722-4726, Taipei, Taiwan: October 8-11, 2006
[5] Luan Ou and Hong Peng, "Knowledge and Process Based Decision
Support in Business Intelligence System," Proceedings of the First
International Multi-Symposiums on Computer and Computational
Sciences (IMSCCS), 2003
[6] B. Delibasic, et al., "Towards Knowledge in Business Intelligence,"
8th Balkan Conference on Operational Research, pp. 97-10, Belgrade,
Zlatibor, September 14-17, 2007.
[7] C. M. Olszak and E. Ziemba, "Approach to Building and
Implementing Business Intelligence Systems," Interdisciplinary
Journal of Information, Knowledge and Management, Vol. 2, pp.
135-148, 2007
[8] W. Zhong, et al., "A Framework of Applying BI to Social Security
Systems," Proceeding of the International Conference on Intelligent
Computation Technology and Automation, pp. 189-193, 2008
doc_927725572.pdf
Business Intelligence Modeling A Case Study of Disaster Management Organization in Pakistan
Business Intelligence Modeling: A Case Study of Disaster Management
Organization in Pakistan
Sohail Asghar
Department of Computer Science
Mohammad Ali Jinnah University,
Islamabad, Pakistan
e-mail:
[email protected]
Simon Fong
Department of Computer and
Information Science
University of Macau
Macau SAR
e-mail:
[email protected]
Touqeer Hussain
Shaheed Zulfikar Ali Bhutto
Institute of Science and
Technology,
Islamabad, Pakistan
e-mail:
[email protected]
Abstract—Acquiring useful Business Intelligence (BI) for high-
qualitative decision making is a challenging task in today’s
dynamic business environment. Powerful tools are available
such as ETL, data-warehouse, OLAP, data-mining and
visualization. However, companies are lack of standardization
of procedures in cultivating BI for improving the performance
of the organization at all levels. In this paper, we developed a
model that link dimensions of BI and processes together, for
providing a good decision support, for disaster management
organization in Pakistan as a case study. The model is
implemented and validated using Oracle BI tools and
techniques. The significance of this research work is the BI
model proposed which provides exploratory abilities on the
data, and linkages among BI processes from the conceptual BI
dimensions.
Keywords-Business Intelligence dimension, Data Mining,
OLAP, Data Warehouse, and Business Functionaity
I. INTRODUCTION
BI is recognized as an increasingly important support for
business decision making [1] in today’s emerging business
environment, where a huge amount of data is growing fast
and scattered around. Most organizations nowadays face the
challenges of processing and analyzing enormous amount of
data [4] for real-time BI. To tackle such problems, data are
progressively transformed into useful information and
knowledge management techniques are implemented to
manage the information and to support decision making [5].
In assimilation to above BI also supports different
enterprises by utilizing its tools and techniques [6] in order to
support them in carrying out different processes and handle
the processes according to different tasks. In regard of this
BI is identified as an amalgamation of reporting, data mining
and online analytical processing applications. BI supposed to
provide access to data that has been cleaned and integrated
so that they can be analyzed, manipulated, transformed, and
combined to discover correlations, trends, and patterns that
offer new insights, aid in decision-making. However,
enterprises adopt different practices for BI by using a mix of
ETL, data warehouse, OLAP, data mining and decision
models [1][4][8]. Data warehouse is the core of BI system
which stores aggregated and historical data [1]. It is loaded
from many operational data sources i.e. MIS, CRM, ERP and
other legacy systems [1][4][6], by some automated
procedure called ETL (Extract, Transform and Load) [7]. On
top of that, OLAP (Online Analytical Processing) provides
analytical views of the stored data in the warehouse in the
form of star or snowflake schema [1][3][4]. On the other
hand, Data Mining helps to find hidden patterns from the
data through different algorithms (association rule mining,
decision tree, and clustering etc).
With fast growth of business data in both volume and
variety, amalgamation of data from different sources by
different autonomous processes into the corresponding BI
dimensions is a challenge. To cope with that, we have
proposed a model which essentially provides an exploration
and linkage among the BI processes and its dimensions for
good decision making. In this paper we describe a detailed
validation and implementation of the model through a real-
life case study. Some key tools and techniques used include
Oracle 11g, Oracle analytical work manager (for OLAP) of
data warehousing for its multidimensional cube generation.
For data representation in cubes as a user interface, some
operations have been carried out like slicing, dicing, rollup,
drill down and that of pivoting. The proposed model is
composed of two main parts: BI dimension and BI process.
BI dimension embraces the management concepts of
knowledge, functionality, technology, business and
organization; whereas BI process includes the technical
activities of operational data sources, ETL, data warehouse,
OLAPS, data mining and visualization tools.
Consequently, we have linked the BI dimension with the
BI process in our model, which is essential during the life
cycle of BI system development. As a result of the
integration it is possible for us to implement efficient
architecture by quick prototyping. The developed prototype
architecture is then tested with real life data. The
experimental results are validated and we highlight the
benefits attained.
The remaining portions of this paper are organized as
follows: Section 2 presents a literature review of current
research efforts by various researchers in the area of BI
models and methodology. We show our proposed
architecture of BI system in Section 3. Section 4 discusses
the validation and evaluation of experimental results. Finally,
Section 5 provides a conclusion and possible future research.
II. LITERATURE REVIEW
In 2003, Olszak and Ziemba [1] suggested an integrated
approach to build and implement business intelligence
solution. This integrated approach composes of four basic
dimensions i.e. business, function, technology and
organization. According to the authors an adequate approach
is required to design, implement and use the BI system,
keeping in mind the four dimensions where BI systems are
being initiated. The same approach has been suggested by
the [2] and [7]. Our approach is similar with respect to their
proposed BI dimension. But we are considering also the
knowledge as an additional dimension. We explored these
dimensions and linked them with BI creation,
implementation and using the process as shown in Fig. 1. We
implemented this approach in practical domain and
developed a BI application whereas [1] only suggested the
implementation approach.
Later in 2004, Simmer [2] proposed a stakeholder model
of business intelligence which was closely related to the
model presented in [1]. According to the author for the
purpose of expansion in the organization performance, this
model is potentially useful for design, diagnosis, and
enhancement of BI because it integrates stakeholder system
with technical and human knowledge systems. The
stakeholder model of BI is mainly divided into two parts:
external environment and internal environment. Whereas the
external environment covers the institutional, extra-
institutional and industry environment and internal consists
of business practices, knowledge base, knowledge source,
value creation, and strategic attributes. However, the
connection of stakeholder model with respect to our
proposed model is similar based upon the BI dimension. The
external factors of stakeholder model which we have covered
into our organizational dimension and other dimension are
covers into internal factors. An important feature of our
model is that it is quite generic because we have also focused
on the BI process and extensively explored it which is the
limitation of the model presented in [1] and [2].
Xu et al [3] presented BI applications infrastructure for
the life insurance industry. They have demonstrated that how
an innovative BI infrastructure and application could
effectively address major business challenges. Here arises a
question of how this will help us to achieve operation
excellence and business impact. The major parts of the
infrastructure are meta-DBs, which are operational data
sources. The second part is a BI application that includes
data mining, OLAP, data queries and data warehouses. The
last part is BI users including end-users of the application,
based on their roles/responsibilities. The presented work is
limited in relation to our proposed model that it cannot
identify the transformation of data from operational data
sources to data warehouse, which is an essential part of the
BI system. Our proposed model however highlights on the
transformation procedure from operation data sources to data
warehouse, which is illustrated in Fig. 1.
In 2006 Zeng et al. [4] claimed that successful
application of BI in an enterprise should be measured
through two reasons. The first is correct, valid, integrated
and in-time data, and the other is the means which
transforms the data into decisions. Both tasks have certain
difficulties. Therefore the authors presented a BI technical
framework which consists of an operational application tier
(legacy data sources), a data acquisition tier (ETL), a data
warehouse tier, BI suite tiers and corporate performance
management tiers. The model is quite similar to our
proposed model except that our model has details in
relationship of BI dimensions and BI processes which
mutually facilitate decision making.
In the fall of 2006, Ou and Peng [5] introduced the
concepts of business process management to the current
business intelligence with the ability of process-driven
decision making. These processes are stored in process
model base and are flexible and reusable. By using case
based reasoning technology, matching processes can be
retrieved and delivered to the decision makers when they
encounter a new problem. The authors tried to address the
situation when the existing data sources could not fulfill the
problem statement of new problem. In this situation the
assistance of knowledge implementation and business
process reuse can attain the decision making efficiency of BI
system. To deal with these issues the author proposed a
knowledge base business intelligence system (KBBIS). With
respect to our proposed model the knowledge dimension also
deals with such types of issues when the existing data
sources are insufficient to fulfill the problem statement.
Delibasic in 2007 [6] presented an abstract level model of
business intelligence system. It proposes to reach for
decision making through knowledge in two ways i.e. with
experts and computational supports. At experts side a
decision maker has support from the GDSS (group decision
supports system) with experts’ collaboration. The main parts
of the model are databases for the computation support,
various knowledge discovery tools (case based reasoning
system, data mining etc.) and visualization techniques. Our
model is supporting these features which are closely related
to the model presented in [6]. It also covers other aspects of
BI system which are not discussed by the author. Our model
covers a comprehensive range of dimensions such as
business, technology, functionality and organization, which
are very important in designing the BI systems.
Olszak and Ziemba in 2007 [7] introduced a
methodology of BI system creation and implementation for
organizations. This approach is focused on objectives and
functional areas of BI in organizations. According to the
authors, the BI has two major stages that are of interactive
nature, (1) BI creation and (2) BI consumption. The creation
stage involves a series of tools and technologies, which
include ETL, data warehouse, OLAP, data mining and
different presentation tools. Consumption stage involves the
fundamental changes in a particular enterprise. The authors
focused on the issue that organizations require some cultural
background to go along with information system and
information technology, when building and implementing a
BI system. This suggested methodology of building and
implementing BI system also need sound business practices
set by the enterprise. Our proposed model does not only
focus on the above listed issue, it actually provides a real
time application in the disaster management domain and
provides a better decision-making environment.
Zhong et al. in 2008 [8] applied BI into a social security
system. It provides business analysis to the decision makers.
Authors proposed a three-dimensional framework of BI
which is based on a three-level screening funnel. It refers to
the dimension of the user roles, the dimension of instruments
and the dimension of contents. In terms of three-level
screening funnel it advocates that it is necessary to screen by
industry positioning, positioning of value chains and
positioning of the stage for expanding users. This idea leads
to the best match of BI applied, namely carrying out concrete
operations. The role dimension is made up of three kind of
personnel namely, business personnel, technical personnel,
and business technical personnel. The content dimension
reflects application, inclusive of deductive application,
inductive application and the application for controlling and
developing the instrument dimension.
III. PROPOSED BUSINESS INTELLIGENCE MODEL
The proposed architecture is divided into two main parts
i.e. BI dimension and BI process. BI dimension includes the
concept of knowledge, functionality, technology, business
and organization whereas the BI process includes the
activities of operational data sources, ETL, data warehouse,
OLAPS, data mining and visualization tools. As a holistic
approach, we linked the BI dimension with the BI process as
a whole model which is essential during the complete life
cycle of BI system development.
A. BI Knowledge aspect
According to [1] knowledge is asserted for enterprise
which is applied in all key business processes. It also
constitutes a prerequisite for the development of new
products and technologies, volume of scales, reaching new
customers and maintaining relations with existing customers.
The originating source of knowledge in an enterprise
includes: their information systems, internal documentation,
media press, reports, domestic and foreign statistics, internet,
corporate databases, customers, suppliers, business partners,
and their employees [1][2]. However, knowledge can be
classified into following categories i.e. procedural,
descriptive, semantic, episodic, explicit and tacit knowledge
[2][6]. As a result, it is very important for an enterprise to
analyze the source of knowledge, data sources, knowledge
type and solution for the problem when existing data sources
are unable fulfill the requirement for the BI system.
B. BI Functional aspect
The second important dimension which is linked with the
BI process is the functional aspect. According to [1] BI
standard should be analyzed taking into consideration of all
the benefits which are likely to be generated in an enterprise.
Following are the functional aspect that is considered during
BI process [1]:
• Strategic planning included:
o Different variants in the development of an
organization
o Relation of an enterprise strategy, mission, goals
and tasks
o Identification of bottlenecks to be tackled
o Providing information on the enterprise
environment & market trends
• Improving relation with customers
• Analyzing profitability of products
• Analyzing the internal processes and operational
efficacy
• Controlling and managing accounting
Figure 1. Business Intelligence model.
C. BI Technology aspect
The third aspect of BI system is technology. According
to [1][4] different world-wide leading BI vendors provide
end-to-end advanced enterprise platforms; BI solution
primarily involves method of knowledge creation, source of
knowledge and information technology tools. However,
integration between world-class databases management,
analysis server, enterprise sources and flexible, easy-to-use
front-end applications make the power of BI accessible to
users at all levels of the organization and realize benefits in
some key access.
More and More enterprise solution and platform for BI
have been developed. Some commercial solution include,
just to name a few, Oracle databases, IBM DB2, Microsoft
SQL Server [1][4][7], with the BI tools Microsoft analysis
server, Oracle Analytical Work Manager (AWM), Oracle
Warehouse Builder [1], Oracle BIEE (Business Intelligence
Enterprise Edition), NCR teradata warehouse, Hyperion,
SAS, Cognos, Business Objects [7], Open source project
Pentaho, and MS Miner [4].
However, selecting appropriate BI tools from the above
list vender is not an easy task. The authors in [7] suggest a
set of criteria that should be taken into consideration during
the selection. The criteria are based on the following factors
that are being analyzed: functionality, complexity of
solution, compatibility, and organizational functional needs.
However, it is also important that these tools meet the
organization expectations in a foreseeable future.
D. BI Business aspect
The fourth important aspect of BI system is business.
According to [2] the increasing use of business practices has
given rise to spiraling demands for information. BI is able to
codify and share these demands through multiple internal
and external environments. In turn technical development
within BI has enabled the increased diffusion of these
business practices.
E. BI Organization aspect
The last but not the least important aspect of BI system is
organization. It is imperative to understand the business,
goals of an enterprise before implementing the BI solution
[1] because the present and future demands for knowledge in
an enterprise are based on these goals. Next tasks are to
analyze and design key processes, positions, responsibilities,
and stream knowledge flows.
IV. VALIDATION AND EVALUATION
This section is presented in five phases: in the first phase
we explain the implementation model details and elaborate
the different parts of the model; in the second phase we
enlighten the data warehouse schema, its development in
Oracle Warehouse Builder (OWB) and also the ETL process;
in the third phase we clarify the implementation details of a
case study with Oracle BIEE tool; in the fourth phase we
explain the ad-hoc queries implementation details with
Oracle Discover and Discover Plus (that are embedded in
Oracle Application Server 10g); we explain the integration of
Oracle BI solution with MS-Office which is very useful for
different analysis purposes in the last phase.
The implementation model is depicted in Fig. 2 and it is
divided into four main parts: the first part is operational data
sources which include the three main data sources physical,
financial and training. We had applied ETL process on these
data sources and the lode data into data warehouse shown in
Fig. 3. We mapped them on a logical schema called data
warehouse schema shown in Fig. 4. The schema consists of
three main facts tables and two dimensions which are TIME
and LOCATION respectively.
On the third step of implementation model we had linked
Oracle BIEE, Oracle Discover plus & Desktop, and MS-
Office with warehouse for data presentation. This
presentation comes with web browser and desktop utilities.
Fig. 5 gives a preview of business model development in
BIEE administration tools. This model contains three main
layers, i.e. physical, business model & mapping, and
presentation. Physical layer consists of actual data sources
imported from a data warehouse which is consisting of
dimension and facts tables. Based on this physical data
source we built a business model & mapping layer where we
could logically define a hierarchy of LOCATION and TIME
dimensions that are depicted in Fig. 6. When the hierarchy is
defined, it can be published in the presentation layer.
Figure 2. Business Intelligence Implementation Model.
Figure 3. Data Transformation Services (SQL Loader) wizard view.
Figure 4. Schema Development in Oracle Data Warehouse 11g (OWB)
Screen view.
After completing these steps we established an OC4J
server and named it BI Presentation Services which is a web
based interface shown in Fig. 6. In this interface we
compiled the different answers and save them into
appropriate folders, which in turn can be used as useful
queries presented on the dashboard as shown in Fig. 7
Figure 5. Business Model Developments in BIEE Administration Tool.
Figure 6. BIEE Answer Development wizard.
Figure 7. Real Time Activities Monitoring Dashboard.
Figure 8. Business Model Developments in BIEE Administration Tool.
Figure 9. Integration with MS-Excel 2003 as BI viewer.
The interactive dashboard is very useful for analysis
purposes. Users can roll-up and drill-down any answer
presented on the dashboard and move into the next level of
information which is in deeper details by a given domain. In
our case study, our users can navigate information up to four
levels: Province District Tehsil Union Council.
The ad-hoc query presented in Fig. 8 is built in Oracle
Discover Plus. User can build different cross-tabs and
multiple graphs from which BI reports can be printed.
Fig. 9 and Fig. 10 demostrate the integration service of
Oracle BI tools with MS-Excel and MS-Power Point
respectively. It is effective for the users that they can easily
call a query from the Oracle dashboard and display them in
BI viewers with the data in the MS-Excel formats.
Figure 10. Integration with MS-Power Point 2003 as BI.
V. CONCLUSION AND FUTURE WORK
In this project we linked BI dimension with BI process
which is a primary source of BI solution creation in any
organization. We have devised a architecture for the BI
system and made it more useful with this mapping of BI
process and dimension. Furthermore, the proposed
architecture along with its components is discussed,
highlighting their limitation and advantage. A prototype has
been developed and experiments are performed on a case
study of disaster management organization in Pakistan.
It is observed that our architecture is possible to
implement and manage. We experimented with several sets
of data, the results show that they are quick and efficient
decision-making can performed over the BI system.
Currently we are trying to explore the relation of
enhanced OLAP, different data mining techniques, real time
data warehousing and DSS. We are now in exploratory phase
of how these technologies contribute the BI system for
increasing its performance and visualization. The ultimate
goal is to develop a better business decision-making
environment.
REFERENCES
[1] C. M. Olszak and E. Ziemba, "Business Intelligence as a key to
Management of an Enterprise," Proceedings of Information Science
and IT Education Conference, pp. 855-863, June 2003
[2] Claire A. Simmers,"A Stakeholder Model of Business Intelligence,"
Proceedings of the 37th Hawaii International Conference on System
Sciences, pp. 1-9, 2004
[3] Z. Xu, M. Zhang and X. Jiang, "Business Intelligence - A Case Study
in Life Insurance Industry," Proceeding of the International
Conference on e-Business Engineering (ICEBE), pp. 129, 2005
[4] Li Zeng, et al, "Techniques, Process and Enterprise Solutions of
Business Intelligence," IEEE Conference on System, Man, and
Cybernetics, pp. 4722-4726, Taipei, Taiwan: October 8-11, 2006
[5] Luan Ou and Hong Peng, "Knowledge and Process Based Decision
Support in Business Intelligence System," Proceedings of the First
International Multi-Symposiums on Computer and Computational
Sciences (IMSCCS), 2003
[6] B. Delibasic, et al., "Towards Knowledge in Business Intelligence,"
8th Balkan Conference on Operational Research, pp. 97-10, Belgrade,
Zlatibor, September 14-17, 2007.
[7] C. M. Olszak and E. Ziemba, "Approach to Building and
Implementing Business Intelligence Systems," Interdisciplinary
Journal of Information, Knowledge and Management, Vol. 2, pp.
135-148, 2007
[8] W. Zhong, et al., "A Framework of Applying BI to Social Security
Systems," Proceeding of the International Conference on Intelligent
Computation Technology and Automation, pp. 189-193, 2008
doc_927725572.pdf