DATA INTEGRATION ISSUES FOR BUSINESS INTELLIGENCE
INTEGRATED ENTERPRISE INFORMATION SYSTEMS
Pierre F. Tiako
Center for Information Technology Research, Langston University, Oklahoma, USA
Keywords: Data Integration, Enterprise Information System, Business Intelligence, Distribution, Web Technology.
Abstract Business Intelligence (BI) provides the ability to access any type of data inside or across enterprises and to
analyze and present them as usable information. To work on business intelligence, an enterprise has to deal
with important problems relating to both (1) Data integration and (2) Analysis and presentation of data for
strategic decision-making. No matter what the application, the need for business intelligence applies
universally. This position paper focuses on data integration issues for business intelligence integrated
Enterprise Information Systems (EIS).
1 INTRODUCTION
Today’s enterprises are judged not only on the
quality of their products and services, but on how
well they share information with customers,
suppliers, shippers and banks in the Supply Chain.
The more widely available information is throughout
the enterprise, the more valuable it becomes. The
information actors in the Supply Chain are not in the
same location, but are distributed among them. Each
actor has its own data representations and
presentation mechanisms. This makes sharing and
exchanging information across partners extremely
difficult. The more integrated an enterprise becomes
the easier it is for customers and other partners to
access data they need. They are then empowered to
make their best decisions to buy or sell.
Several research and industrial contributions
(Builders, 1999; Acta, 1999; Enosys, 2002)
introduced the concept of business intelligence to
refer to an Internet-driven organization that uses
information in an intelligent fashion. Business
intelligence leverages the ubiquitous Internet
network to deploy information consistently and
accurately across departments and business units of
the same enterprise internally or among autonomous
partners externally involved in the Supply Chain.
Business intelligence doesn’t replace e-business;
rather, it is a critical part of an e-business strategy.
This report progress focuses on critical data
integration issues for business intelligence integrated
EISs. In continuation, this paper is organized as
follows. Section 2 is devoted to present concepts of
data integration for business intelligence. Section 3
presents critical issues for designing integrated EISs
for business intelligence. Section 4 includes
discussion on related work and perspectives.
2 BUSINESS INTELLIGENCE AND
DATA INTEGRATION
Business intelligence can be regarded as use of
information for strategic decision-making. From this
point of view, business intelligence mainly deals
with data integration and strategic decision-making
processes. BI supports the following cognitive
phases: (1) Exploration of the world of information,
(2) Interrogation of the base of information, (3)
Analysis of the base of information, (4) Annotation
based on the individual preferences and discoveries
(David and Thierry, 2003).
2.1 Business Intelligence
Business intelligence provides the ability to access
any type of data inside or across enterprises and to
analyze and present them as usable information. To
work on business intelligence, an enterprise has to
deal with important problems relating to both (1)
Data integration and (2) Analysis and presentation of
data. No matter what the application, the need for
business intelligence applies universally. For e-
commerce applications targeting the mass consumer
market, the most important aspect is intelligence
366
F. Tiako P. (2005).
DATA INTEGRATION ISSUES FOR BUSINESS INTELLIGENCE INTEGRATED ENTERPRISE INFORMATION SYSTEMS.
In Proceedings of the Seventh International Conference on Enterprise Information Systems, pages 366-369
DOI: 10.5220/0002538303660369
Copyright
c
SciTePress
about the likes, dislikes, demographics and behavior
patterns of customers and prospects; just as critical
for business-to-business vendors is analysis about
partners, suppliers, and manufacturing and
distribution activities. Without business intelligence,
competitive e-businesses are flying blind.
Integration challenge (Kiper, 1994; ETI, 2002)
requires different approaches to the problem that
supplements the traditional messaging solutions
offered by Enterprise Application Integration (EAI).
EAIs propose various architectural approaches to
integration, but most have a wide range of interfaces
and work by simply broadcasting transaction
information to other connected systems.
2.2 Hierarchical view of Integrated
EISs for Business Intelligence
Each project related to tools integration (Brown,
1992) must be categorized according to the five
dimensions of integration proposed by Wasserman
(1990), Thomas and Nejmeh (1992). The
dimensions are presentation, platform, control data,
and process integration. For software environments
resulting from integrated project support
environment projects, Brown (1992) considers three
aspects of integration (see figure 1): presentation,
control and data integration. Presentation integration
deals with the way users interact with EISs using
services of user interface manager. Control
integration aims at coordinating tools execution
(Brown and Wallnau, 1996). A tool invokes another
for realizing an action, rather than implementing that
action. Data integration deals with the problems of
sharing or exchanging data between tools.
2.3 Research and Industrial
Approaches
Industrial solutions dealing with data integration
favorable to business intelligence (see table 1) can
be grouped into the following categories: (1) EAI
systems, which are vital to the creation of
collaborative e-business applications because they
enable the integrity of data and business rules to be
preserved; (2) Business-to-business integration
(B2Bi) systems that focus on extending EAI
between companies across the Internet (they
typically do not offer in-depth data integration
capabilities beyond Web-scraping technologies); and
(3) Enterprise Information Integration (EII) that can
be distinguished from the other by its underlying
technology that provides real-time, seamless access
to any data, anywhere, for any application.
Depending on each enterprise’s unique situation,
EII can be applied to perform rapid, high-level data
integration either as a stand alone solution or within
the framework of a larger integration project.
Because EII technologies work over standard web
connections to access an array of data sources
(Dossick and Kaiser, 1996) they can be invoked as a
web service or embedded within an integration
application rapidly and easily.
Category Definition Function Technology
(Industry)
Enterprise
Application
Integration
(EAI)
Systems
Systems that
link disparate
enterprise
applications
Integrate
applications at
business
process level
with focus on
transaction.
RMI Applet
(Java Soft)
OLE
(Microsoft)
ODBC JDBC
APIs
(Research)
Control
Integration
(SoftBench,
2002)
Field(Reiss,
1994)
Approaches that
control tool
during
integration
process
Handle
integration
messages
exchanges
between EIS
CORBA
(OMG)
DCE (OSF)
SOM/DS0M
(IBM)
(SOM, 1994)
(Industry)
Business-to-
business
integration
(B2Bi)
Systems
Systems that
automate point-
to-point trading
transactions
among partners
Help migrate to
the web without
disrupting
existing
business
practices
HTML
CORBA
(OMG)
(Research)
Data
Integration
(Boudjlida,1
995)
Contributions
that deal with
data integration
during the
integration
process
Support
semantic and
syntactic
information
relating to data
during
integration
CORBA
(OMG)
DCE (OSF)
SOM/DS0M
(IBM)
(Industry)
Enterprise
Information
Integration
(EII)
Systems
Systems that
access,
integrate, and
unite data from
multiple sources
Provide real-
time data access
within and
through
corporate
firewalls.
XML
CORBA
(OMG)
Table 1: Existing Technologies for integrated
EIS
End User
Presentation
Control
User Interface
Manageme
Message
Repository
Protocol
Figure 1: Hierarchical view of an EIS's architecture.
DATA INTEGRATION ISSUES FOR BUSINESS INTELLIGENCE INTEGRATED ENTERPRISE INFORMATION
SYSTEMS
367
The same research contribution dealing with data
integration in favor of business intelligence can be
group into the following categories: (1) data
integration, which is concerned with data
representation and naming. Significant contribution
carried out in data integration approaches proposes a
dynamic approach based on an abstract model,
which captures data semantics and constraints. The
model includes a set of constructors, which allow
defining complex structures and a set of operators
that determine the transformations that could be
applied to change one type to another. (2) Control
integration, which is concerned with logic or
structure of programming model. Control also
implies that a tool can decide whether and how
much to share operations it supports and data it
manages. Thus a tool could be added or removed
from an EIS; it could participate in more than one
interoperating EIS.
Despite the number of research and industrial
approaches to address, they are almost partial
because each of them is expressively aimed at
addressing a particular aspect of integration. Again
no matter what the application, the need for business
intelligence applies universally. The idea of data
integration for business intelligence is to provide a
more general architecture supporting tool
integration, for we need to combine a set of
mechanisms in developing a complete and coherent
system, as presented below.
3 CRITICAL ISSUES FOR
DESIGNING INTEGRATED EII
FOR BUSINESS INTELLIGENCE
The critical issues for designing Integrated
Enterprise Information Systems for Business
Intelligence (IEISBI) are listed below. IEISBI have
to address these issues in order to be competitive.
Recent research, as well as new products, focuses
on Web Services as a means to integrate data.
However, Web Services are new technology and all
the requirements discussed above and issues listed
below still apply. For instance, semantic Web
Services (WSMO, 2004) addresses the requirements
and issues taking all aspects into consideration.
(a) Communication. IEISBI technology must
provide reliable and secure communication
between back end applications in order to
ensure overall consistency
(b) Distribution. IEISBI applications are
distributed in the sense that each has its own
separate storage or database management
systems with each separately controlled and
managed
(c) Heterogeneity. IEISBI applications are
implemented based on their own data model and
due to the particular management focus the data
models differ in general not complying with a
common standard.
(d) Mediation. Due to heterogeneity of IEISBI
applications, objects are represented in different
data models that have to be mediated if objects
are sent between applications.
(e) Process management. Multi-step business
processes across IEISBI applications require
process management (Tiako, 1998) to
implement the particular invocation sequence
logic.
(f) Reliability. The integration of IEISBI
applications must be reliable in order to neither
loose data nor accidentally introduce data by
retry logic in case of failures
(g) Replication. Replication is a mechanism that
ensures that changes of an object are
automatically propagated to duplicates of this
object. The various representations act as if they
are one single representation
(h) Security. Communication of data between
IEISBI applications through EIS technology
must ensure security to avoid improper access.
4 RELATED WORK AND
PERSPECTIVES
Here we briefly present some EIS tools available in
the market and susceptible to favor business
intelligence by data integration. WebFOCUS
(Builders, 1999) connects core business systems,
enabling data to flow freely from one to another.
ETI (2002) automate the process of data collection,
transformation, and migration between incompatible
systems in heterogeneous computing EISs.
ActaWorks (Acta, 1999) solutions connect ERP
systems to business to business web applications, to
business to commerce web applications and to
direct-to-the-web data access applications.
Informatica (2000) outlines the current state of
convergence of e-business and business intelligence,
and shows how its data integration solutions support
the key infrastructure demands of e-businesses.
Enosys (2002) shows how XML presents significant
advantages as the data representation language for
modern e-business applications and addresses their
needs better.
ICEIS 2005 - DATABASES AND INFORMATION SYSTEMS INTEGRATION
368
The emerging of the Internet has accentuated the
need for supporting a broad diffusion of information
and the openness of EISs. To meet these needs, user-
friendly interfaces must be based on HTTP protocol
(Riva and Ramoni, 1996) and HTML. So, user
interfaces should benefit from standards of scripting
programming languages such as JavaScript
(Flanagan, 2002) for increasing the interactivity with
users.
The distribution of components of an EIS can be
realized by coupling Web technologies and
distributed object approaches (Brose et al., 2002)
such as CORBA or COM, including a Java
implementation of its specifications. The resulting
platform built on top of this kind of association are
opened, and characterized by the following
properties:
(a) Neutrality of the platform. Means independent
of the underlying operating system and
hardware architecture.
(b) Portability of components. Components of an
EIS can be moved from one platform to another,
with little effort, by adapting the new
component for its integration in an already
existing EIS.
(c) Distribution. The user interface managed by
the Web browser is completely separated from
the server that supports the processing and the
storage of data. Interoperable tools which make
up an EIS are installed on different computers.
This way of distributing components of an EIS, by
using the Web as support, is adopted by different
approaches, which are grouped here in two
categories. The difference between these proposals
lies in methods the Web accesses legacy server
applications and their data. Data integration issue for
business intelligence presented in this paper differs
from the above solutions by its openness. The future
of this work will study the issues presented above.
REFERENCES
David, A., and O., Thierry, 2004. EQuA2te Architecture
and its application in Business Intelligence. Business
Intelligence: Research & Applications Conference -
IERA 2003 - Nancy, France.
Acta, 1999. ActaWorks™ RealTime Technical Overview:
A Data Integration Platform For eCommerce And
Business Intelligence. A White Paper - Acta
Technology. http://www.acta.com.
Boudjlida, N., and O., Perrin, 1995. An Open Approach
for Data Integration. In 2nd Int. Conf. on Object-
Oriented Information Systems (OOIS'95), Dublin.
Brose, G., et al., 2002. JacORB Programming Guide
1.4.1. Freie Universität Berlin and Xtradyne
Technologies AG.
Brown, A., W., 1992. Control Integration through
Message Passing in a Software Development
Environment. Technical Report, CMU/SEI-92-TR-35.
Brown, A., W., and K., C., Wallnau, 1996. Engineering of
Component-Based Systems. Component-Based
Software Engineering: Selected Papers from the
Software Engineering Institute. IEEE CS Press.
Builders, 1999. The Critical Role of Business Intelligence
in e-Business. A White Paper - Information Builders
Dossick, S., E., and G., E., Kaiser, 1996. WWW Access to
Legacy Client/Server Applications. Computer
Networks and ISDN Systems, The International
Journal of Computer and Telecommunications
Networking, 28(7-11):931-940, Elsevier Science B.V.
Enosys, 2002. Enterprise Information Integration:
Integrating the Real-time Enterprise. Business White
Paper 8 Enosys Software Confidential.
ETI, 2002:. Data integration management for the
enterprise. The ETI solution TM version 5.
Evolutionary Technologies International.
http://www.eti.com.
Flanagan D., 2002. JavaScript: The Definitive Guide, 4
th
Edition. Sebastopol, Ca.: O’Reilly.
Informatica, 2000. Business Intelligence and the New
World of E-business. A White Paper WP1049-B.
Informatica. www.informatica.com
Kiper, J., D., 1994. A Framework for Characterization of
the Degree of Integration of Software Tools. Journal
of System Integration, No. 4, pp. 5-32.
Reiss, S., P., 1994. The FIELD Programming
Environment. Kluwer Academic Publishers.
Riva, A., and M., Ramoni, 1996. LispWeb: a specialized
HTTP server for distributed AI applications. Computer
Networks and ISDN Systems, 28(711):953-961, 1996.
SoftBench, 2002. Integrating with HP SoftBench. HP-UX
STK version 1.8.1. Hewlett-Packard.
SOM, 1994. The System Object Model (SOM) and the
Component Object Model (COM): A comparison of
technologies from a developer's perspective. IBM
Corporation, Object Technology Products Group
Thomas, I., and B., A., Nejmeh, 1992. Definitions of Tool
Integration for environments. IEEE Software Vol. 9,
N° 3, pp 29-35.
Tiako, P., F., 1998. Modeling the Federation of Process
Sensitive Engineering Environments: Basics Concepts
and Perspectives. In Software Process Technology (V.
Gruhn, Eds) - LNCS Vol. 1487, Springer-Verlag.
Wasserman, A., I., 1990. Tool Integration in Software
Engineering Environments. In Software Engineering
Environments. International Workshop on
Environments, LNCS N° 467, pp.137-149.
WSMO, 2004. Web Service Modeling Language,
www.wsmo.org.
DATA INTEGRATION ISSUES FOR BUSINESS INTELLIGENCE INTEGRATED ENTERPRISE INFORMATION
SYSTEMS
369