Вы находитесь на странице: 1из 14

DEFINITIONS OF BI

Business intelligence (BI) is an umbrella term that combines architectures, tools, databases, analytical tools, applications, and
methodologies (see Raisinghani, 2004). It is, like DSS, a content-free expression, so it means different things to different
people. Part of the confusion about BI lies in the flurry of acronyms and buzzwords that are associated with BI (e.g.,
business performance management [BPM]). BFs major objective is to enable interactive access (sometimes in real-time) to
data, to enable manipulation of data, and to give business managers and analysts the ability to conduct appropriate analysis.
By analyzing historical and current data, situations, and performances, decision makers get valuable insights that enable
them to make more informed and better decisions (see Zaman, 2005). The process of BI is based on the transformation of
data to information, then to decisions, and finally to actions. (Details are provided in Chapters 5-7.)

A BRIEF HISTORY OF BI

The term BI was coined by the Gartner Group in the mid-1990s. However, the concept is much older; it has its roots in the
MIS reporting systems of the 1970s. During that period, reporting systems were static, two dimensional, and had no
analytical capabilities. In the early 1980s, the concept of executive information systems (EIS) emerged. This concept
expanded the computerized support to top-level managers and executives. Some of the capabilities introduced were dynamic
multidimensional (ad hoc or on-demand) reporting, forecasting and prediction, trend analysis, drill-down to details, status
access, and critical success factors (see Chapter 6). These features appeared in dozens of commercial products until the mid-
1990s. Then the same capabilities and some new ones appeared under the name BI. Today, a good Bl-based enterprise
information system contains all the information executives need. So, the original concept of EIS was transformed into BI. By
2005, BI systems started to include artificial intelligence capabilities as well as powerful analytical capabilities. Figure 1.6
illustrates the various tools and techniques that may be included in a BI system. It illustrates the evolution of BI as well. The
tools shown in Figure 1.6 provide the capabilities of BI. The most sophisticated BI products include most of these capabili-
ties; others specialize in only some of them. For further details, see Zaman (2005) and Raisinghani (2004).

THE ARCHITECTURE OF BI

As discussed in the opening vignette, a BI system has four major components: a data warehouse, with its source data;
business analytics, a collection of tools for manipulating, mining, and analyzing the data in the data warehouse; business
performance management (BPM) for monitoring and analyzing performance; and a user interface (e.g., a dashboard). The
relationship among these components is illustrated in Figure 1.7.
Notice that the data warehousing environment is mainly the responsibility of technical staff, while the analytical
environment (also known as business analytics) is the realm of business users. Any user can connect to the system via the
user interface, such as a browser, and top managers may use the BPM component and also a dashboard.

Some business analytics and user interface tools are introduced briefly in Section 1.9 and in Chapter 6. However, one set of
tools, intelligent systems (see Chapters 12-14), can be viewed as a futuristic component of BI. According to Zaman (2005),
intelligent systems may cause the field's name to be changed to artificial business intelligence.

Data Warehousing

The data warehouse and its variants are the cornerstone of any medium-to-large BI system. Originally, the data warehouse
included only historical data that were organized and summarized, so end users could easily view or manipulate data and
information. Today, some data warehouses include current data as well, so they can provide real-time decision support

WHY WE USE COMPUTERIZED DECISION SUPPORT SYSTEMS

Today's computerized systems possess capabilities that can facilitate decision support in a number of ways, including the
following:

•Speedy computations. A computer enables the decision maker to perform many computations quickly and at a low cost.
Timely decisions are critical in many situations, ranging from a physician in an emergency room to a stock trader on the
trading floor. With a computer, thousands of alternatives can be evaluated in seconds. Furthermore, the benefits-to-cost ratio
of computers and the speed of executions are constantly increasing.

•Improved communication and collaboration. Many decisions are made today


by groups whose members may be in different locations. Groups can collaborate and communicate readily by using Web-
based tools. Collaboration is especially important along the supply chain, where partners—all the way from vendors to
customers—must share information (recall the Toyota TLS case discussed earlier in this chapter; also see Chapters 10 and
17).

•Increased productivity of group members. Assembling a group of decision makers, especially experts, in one place can be
costly. Computerized support can improve the collaboration process of a group and enable its members to be at different
locations (saving travel costs). In addition, computerized support

can increase the productivity of staff support (e.g., financial and legal analysts). Decision makers can also increase their
productivity by using software optimization tools that help determine the best way to run a business (see Chapter 4).

•Improved data management. Many decisions involve complex computations. Data for these can be stored in different
databases anywhere in the organization and even possibly at Web sites outside the organization. The data may include text,
sound, graphics, and video, and they can be in foreign languages. It may be necessary to transmit data quickly from distant
locations. Computers can search, store, and transmit needed data quickly, economically, securely, and transparently.

•Managing giant data warehouses. Large data warehouses, like the one operated by Wal-Mart, contain terabytes and even
petabytes of data. Computers can provide extremely great storage capability for any type of digital information, and this
information can be accessed and searched very rapidly. Special methods, including parallel computing, are available to
organize, search, and mine the data. The costs related to data warehousing are declining.

•Quality support. Computers can improve the quality of decisions made. For example, more data can be accessed, more
alternatives can be evaluated, forecasts can be improved, risk analysis can be performed quickly, and the views of experts
(some of whom are in remote locations) can be collected quickly and at a reduced cost. Expertise can even be derived
directly from a computer system using artificial intelligence methods (discussed in Chapter 12). With computers, decision
makers can perform complex simulations, check many possible scenarios, and assess diverse impacts quickly and
economically.

•Agility support. Competition today is based not just on price but also on quality, timeliness, customization of products, and
customer support. In addition, organizations must be able to frequently and rapidly change their mode of operation,
reengineer processes and structures, empower employees, and innovate in order to adapt to their changing environments.
Decision support technologies such as intelligent systems can empower people by allowing them to make good decisions
quickly, even if they lack some knowledge.

•Overcoming cognitive limits in processing and storing information. According to Simon (1977), the human mind has only a
limited ability to process and store information. People sometimes find it difficult to recall and use information in an error-
free fashion due to their cognitive limits. The term cognitive limits indicates that an individual's problem-solving capability is
limited when a wide range of diverse information and knowledge is required. Computerized systems enable people to
overcome their cognitive limits by quickly accessing and processing vast amounts of stored information (see Chapter 2).

•Using the Web. Since the development of the Internet and Web servers and tools, there have been dramatic changes in how
decision makers are supported. Most important, the Web provides (1) access to a vast body of data, information, and
knowledge available around the world; (2) a common, user-friendly graphical user interface (GUI) that is easy to learn to use
and readily available; (3) the ability to effectively collaborate with remote partners; and (4) the availability of intelligent
search tools that enable managers to find the information they need quickly and inexpensively.

•Anywhere, anytime support. Using wireless technology, managers can access information anytime and from anyplace,
analyze and interpret it, and communicate with those involved.

The User Interface: Dashboards and Other Information Broadcasting Tools Dashboards (which resemble automobile dashboards)
provide a comprehensive visual view of corporate performance measures (also known as key performance indicators),
trends, and exceptions. They integrate information from multiple business areas. Dashboards present graphs that show actual
performance compared to desired metrics; thus, a dashboard presents an at-a-glance view of the health of the organization. In
addition to dashboards, other tools that broadcast information are corporate portals (see Chapter 17), digital cockpits, and
other visualization toois (see Chapter 6). Many visualization tools, ranging from multidimensional cube presentation to
virtual reality, are integral parts of BI systems. Recall that BI emerged from EIS, so many visual aids for executives were
"transformed to BI software (see Chapter 6). Also, technologies such as geographical information systems (GIS) play an
increasing role in decision support.

STYLES OF BI

The architecture of BI depends on its applications. MicroStrategy Corp. distinguishes five styles of BI and offers special
tools for each. The five styles are report delivery and alerting; enterprise reporting (using dashboards and scorecards); cube
analysis (also

Known as slice-and-dice analysis); ad-hoc queries; and statistics and data mining. For further details,

BENEFITS OF BI

As illustrated by the Toyota TLS case, the major benefit of BI to a company is the ability to provide accurate information
when needed, including a real-time view of the corporate performance and its parts. Such information is a must for all types
of decisions, for strategic planning, and even for survival (refer to Application Case 1.5). Eckerson (2003) reported the
results of a survey of 510 corporations that indicates the benefits as viewed by the participants (see Figure 1.8).

Thompson (2004) reported the following to be the major benefits of BI, based on the results of a survey:
•Faster, more accurate reporting (81 percent)

•Improved decision making (78 percent)

•Improved customer service (56 percent)

•Increased revenue (49 percent)


Notice that many of the benefits of BI are intangible. This is why, according to Eckerson (2003), so many executives do not
insist on a rigorous cost-justification for BI projects (see the detailed discussion in Chapter 15).

Thompson (2004) also noted that the most common application areas of BI are general reporting, sales and marketing
analysis, planning and forecasting, financial consolidation, statutory reporting, budgeting, and profitability analysis. An
interesting data mining application using predictive analytics tools (discussed further in Chapters 6 and 7) is illustrated in
Application Case 1.6.

THE DSS-BI CONNECTION

By now, you should be able to see some of the similarities and differences between DSS and BI. First, their architectures are
very similar because BI evolved from DSS. However, BI implies the use of a data warehouse, whereas DSS may or may not
have such a feature. BI is therefore more appropriate for large organizations (because data warehouses are expensive to build
and maintain), but DSS can be appropriate to any type of organization.

Second, most DSS are constructed to directly support specific decision making. BI systems, in general, are geared to provide
accurate and timely information, and they support decision support indirectly. This situation is changing, however, as more
and more decision support tools are being added to BI software packages.

Third, BI has an executive and strategy orientation, especially in its BPM and dashboard components. DSS, on the other
hand, is oriented toward analysts.

Fourth, most BI systems are constructed with commercially available tools and components that are fitted to the needs of
organizations. In building DSS, the interest may be in constructing solutions to very unstructured problems. In such
situations, more programming (e.g., using tools such as Excel) may be needed to customize the solutions.

Fifth, DSS methodologies and even some tools were developed mostly in the academic world. BI methodologies and tools
were developed mostly by software companies. (See Zaman, 2005, for information on how BI has evolved.)

Sixth, many of the tools that BI uses are also considered DSS tools. For example, data mining and predictive analysis are
core tools in both areas.

Although some people equate DSS with BI, these systems are not, at the present, the same. It is interesting to note that some
people believe that DSS is a part of BI— one of its analytical tools. Others think that BI is a special case of DSS that deals
mostly with reporting, communication, and collaboration (a form of data-oriented DSS). Another explanation (Watson,
2005) is that BI is a result of a continuous revolution and, as such, DSS is one of BI's original elements. In this book, we
separate DSS from BI. However, we point to the DSS-BI connection frequently.

Management Support Systems (MSS)

Due to the lack of crisp and universal definitions of DSS and BI, some people refer to DSS and BI, as well as their tools
either independently or in combination, as management support systems (MSS). MSS is a broad enough concept to be viewed as
a technology that supports managerial tasks in general and decision making in particular. In this book, we use MSS when the
nature of the technology involved is not clear, and we use it interchangeably with the combined term DSS/BI.

In addition to the major frameworks of decision support presented so far, we need to look at a new proposed framework —
the work system — which we present next.

By adding noncomputerized means, Alter expanded the landscape of decision support to include nontechnical decision-
improvement interventions and strategies. To cope with the possibility of a huge field with many disciplines, Alter
postulated that decision support may come from the different aspect of work systems. He defined a work system as a system in
which human participants and/or machines perform a business process, using information, technology, and other resources,
to produce products and/or services for internal or external customers. A work system operates within a surrounding
environment, often using shared infrastructure, and sometimes within a conscious strategy for the organization or work
system. Furthermore, Alter postulated that a work system usually has nine elements. Each of these elements can be varied or
modified in order to provide better organizational performance, decision quality, or business process efficiency. The
following are the nine elements, along with some possible sources of improvements:

1.Business process. Variations in the process rationale, sequence of steps, or methods used for performing particular steps

2.Participants. Better training, better skills, higher levels of commitment, or better real-time or delayed feedback

3.Information. Better information quality, information availability, or information presentation


OLAP & OLTP
The term online analytical processing (OLAP) refers to a variety of activities usually performed by end users in online systems.
There is no universal agreement on what activities are considered OLAP. Usually OLAP includes such activities as
generating and answering queries, requesting ad hoc reports and graphs and executing them, conducting traditional or
modern statistical analyses, and building visual presentations. Many people also think of multidimensional analysis and
presentations, EIS/ESS, and data mining as OLAP. Essentially, OLAP products provide modeling, analysis, and
visualization capabilities to large data sets, either to database management systems (DBMS) or, more often, data warehouse
systems, and they also provide a multidimensional conceptual view of the data.

OLAP-VERSUS OLTP

FOR many years, IT concentrated on building mission-critical systems that mainly supported corporate transaction
processing. Such systems must be virtually fault tolerant and provide efficient execution and rapid response. An effective
solution was provided by online transaction processing (OLTP), which centers on routine and repetitive activities, using a
distributed relational database environment. The latest developments in this area are the use of ERP and supply-chain
management (SCM) software for transaction processing tasks, customer relationship management (CRM) applications, and
integration with Web-based technologies and intranets. Many tools were created for developing OLTP applications; the
Informix Dynamic Server (see www-306.ibm.com/ software/data/informix/ids) is an example of one such effective tool.

Access to data is often needed by both OLTP and management support system (MSS) applications. Unfortunately, trying to
serve both types of requests may be problematic (see Gray and Watson, 1998). Therefore, some companies elect to separate
information systems into OLTP types and OLAP types.

OLTP concentrates on processing repetitive transactions in large quantities and conducting simple manipulations. OLAP
involves examining many data items (frequently many thousands or even millions) in complex relationships. In addition to
answering users' queries, OLAP may analyze these relationships and look for patterns, trends, and exceptions. In other
words, OLAP is a direct decision support method.

A typical OLAP query might access a multigigabyte or terabyte, multi-year sales database in order to find all product sales in
each region for each product type. After reviewing the results, an analyst might further refine the query to find sales volume
for each sales channel within a region or within certain product classifications. As a last step, the analyst might want to
perform year-to-year or quarter-to-quarter comparisons for each sales channel. This whole process must be carried out
online, with rapid response time so that the analysis process is undisturbed.

OLAP also differs from OLTP (and also from data mining) in that users can ask specific, open-ended questions. Users,
typically analysts, drive OLAP, whereas data mining looks for relationships, with only some direction from the analyst.
OLAP is generally facilitated by working with the data warehouse (or with data marts or a multidimensional database) and
with a set of OLAP tools. These tools can be query tools, spreadsheets, data mining tools, data visualization tools, and the
like. For a list of OLAP tools, see Alexander (2003), baseline.com, and periodic reviews in the software sections of
PCWeek, DM Review, Intelligent Enterprise, and Software Review. The major vendors of these tools include
BusinessObjects, Computer Associates, Cognos Inc., Geas, Hyperion Software Corp., Informatica Corp., Information
Builders, IBM, Intersolve, M'crosoft, MicroStrategy Corp., Oracle, SAS Institute, Inc., SPSS, and Temtec.

TYPES OF OLAP

The major types of OLAP are:

•Multidimensional OLAP (MOLAP). When OLAP is implemented via a specialized multidimensional database (or data
store), it is called multidimensional OLAP (MOLAP) because it summarizes transactions into multidimensional views (see
Section 6.6) ahead of time. Data are organized into a cube structure that the user can rotate; this is particularly suited for
financial summaries. With MOLAP, queries are fast because the consolidation has already been done.

•Relational OLAP (ROLAP). When an OLAP database is implemented on top of an existing relational database, it is called
relational OLAP (ROLAP). Relational OLAP tools extract data from relational databases. Using complex SQL statements
against relational tables, ROLAP is also able to create multidimensional views on-the-fly. ROLAP tends to be used on data
that has a large number of attributes, where it cannot be easily placed into a cube structure. For example, customer data with
numerous descriptive fields, rather than financial data, are typically ROLAP candidates.

•Database OLAP and Web OLAP (DOLAP and WOLAP). Database OLAP refers to a relational database management
system (RDBMS) that is designed to host OLAP structures and perform OLAP calculations. Web OLAP refers to OLAP data
that is accessible from a Web browser.
•Desktop OLAP. Desktop OLAP involves low-priced, simple OLAP tools that perform local multidimensional analysis and
presentation of data downloaded to client machines from relational or multidimensional databases. Web versions often move
the desktop processing to an intermediate server, which increases the scalability; however, the functionality remains, at best,
comparable to that of the desktop version.

EXAMPLES OF SUCCESSFUL OLAP APPLICATIONS

ING Antai Insurance Co. (Taiwan) uses IBM's OLAP Server to analyze large amounts of data to detect fraudulent claims
and speed up the processing of claims. Now, it takes only a couple of days to analyze data that previously took several
weeks. As the tools and hardware improve, claims can be analyzed faster—even instantaneously. The cost of processing
claims has been greatly reduced. Another example is TCF Bank, which uses OLAP to provide information directly to those
who need it. The bank now understands its customer base more accurately so that it can better target its marketing efforts.

According to Cognos.com (2002), Allied Building Products Corp. has grown its building materials distribution company
by increasing market share and customers.

Allied implemented Cognos Finance, a BI solution from Cognos, in under 90 days. As a result, Allied was able to
standardize data company-wide and automate processing to deliver a single, coordinated view of financial performance.
Cognos Finance allows Allied to reduce manual labor in producing reports and at the same time provides access to budgets,
forecasts, and actuals across all its branch operations. This has enabled an integrated view of information and accelerated
accurate financial reporting.

OLAP TOOLS AND VENDORS

Using SQL and other conventional data access and analysis tools is helpful, but it is not sufficient for OLAP. In OLAP, a
special class of tools is used, including BA front ends, data-access front ends, database front ends, and visual information
access systems. Also, OL*AP methods go well beyond spreadsheets in power and results; they are intended to empower
users. For details, see Perry and Post (2007).

Characteristics of OLAP Tools

OLAP tools have characteristics that distinguish them from MIS tools that are designed to support traditional OLTP
reporting applications. E.F. Codd et al. (1993) succinctly defined the characteristics of OLAP tools in the 12 rules
summarized in Table 6.3. The 12 rules are used for the standardization of multidimensional data modeling, and they define
four types of processing that are performed by analysts in an organization:

1.Categorical analysis is a static analysis based on historical data. It relies on the premise that past performance is an
indicator of the future. This is the primary analysis supported by OLTP transaction-based databases.

2.Exegetical analysis is also based on historical data, and it adds the capability of drill-down analysis. Drill-down analysis is
the ability to query further into data to determine the detail data that were used to determine a derived value.

3.Contemplative analysis allows a user to change a single value to determine its impact.

4.Formulaic analysis permits changes to multiple variables.


Representative OLAP Tools

Vendors in the BI arena are maneuvering to empower end users with the ability to customize analytic applications to meet
evolving business needs. These include, for example, Spotfire's DecisionSite analytics platform (spotfire.com), Business
Objects'

OLAP Product Evaluation Rules: Codd's 1 2 Rules for OLAP

1.Multidimensional conceptual view for formulating queries

2.Transparency to the user

3.Easy accessibility: batch and online access


4.Consistent reporting performance

5.Client/server architecture: the use of distributed resources

6.Generic dimensionality

7.Dynamic sparse matrix handling

8.Multiuser support rather than support for only a single user

9.Unrestricted cross-dimensional operations

10.Intuitive data manipulation

11.Flexible reporting

12.Unlimited dimensions and aggregation level


NEWER APPLICATIONS OF ES

More recent applications of ES include risk management, pension fund advising, business rule automation, automated
market surveillance, and homeland security.

Credit Analysis Systems

ES are developed to support the needs of commercial lending institutions. They can help analyze the credit record of a
customer and assess a proper credit line. Rules in the knowledge base can also help assess the risk and risk-management
policies. These kinds of systems are used in over one-third of the top 100 commercial banks in the United States and Canada.

Pension Fund Advisors

Nestle Foods Corporation has developed an ES that provides information on an employee's pension fund status. The system
maintains an up-to-date knowledge base to give participants advice concerning the impact of regulation changes and
conformance with new standards. A system offered on the Internet at the Pingtung Teacher's College in Southern Taiwan has
functions that allow participants to plan their retirement through what-if analysis that calculates their pension benefits under
different scenarios.

Automated Help Desks

BMC Remedy (remedy.com) offers HelpDesklQ, a rule-based help desk solution for small businesses. This browser-
based tool enables small businesses to deal with customer requests more efficiently. Incoming e-mails automatically pass
into HelpDesklQ's business rule engine. The messages are sent to the proper technician, based on defined priority and status.
The solution assists help desk technicians in resolving problems and tracking issues more effectively. For other examples of
uses of ES with automated help desks, see Online File W12.3.

Homeland Security Systems

PortBlue Corp. (portblue.com/pub/solutions-homeland-security) has developed an ES for homeland security.


It is designed for assessing terrorist threats and provides (1) an assessment of vulnerability to terrorist attack, (2) indicators
of terrorist surveillance activity, and (3) guidance for managing interactions with potential terrorists.

The U.S. Internal Revenue Service uses intelligent systems to detect irregular international financial information and to
block possible money laundering and terrorist financing.

Market Surveillance Systems

The National Association of Security Dealers (NASD) has developed an intelligent surveillance system called Securities
Observation, New Analysis, and Regulations (SONAR) that uses data mining, rule-based inference, knowledge-based data
representation, and NLP to monitor the stock markets and futures markets for suspicious patterns. The system generates 50
to 60 alerts per day for review by several groups of regulatory analysts and investigators ("Automatic Market Surveillance,"
2004).
Business Process Reengineering Systems

Reengineering involves the exploitation of information technology to improve bus ness processes. KBS are used in
analyzing the workflow for business process reena neering. For example, Gensym's System Performance
Analysis Using Real-Ti-d Knowledge-based Simulation (SPARKS) can help model the formal and inforr A
knowledge, skills, and competencies that must be embedded in a reengineered systeiJ SPARKS has three components:
process flow model, resource mode), and work v.-J umes and descriptions.
AREAS FOR ES APPLICATIONS

As indicated in the preceding examples, ES have been applied commercially in a - _oi ber of areas, including the following:

•Finance. Finance ES include insurance evaluation, credit analysis, tax plannir i fraud prevention, financial report analysis,
financial planning, and performar.: i evaluation.

•Data processing. Data processing ES include system planning, equipment se tion, equipment maintenance, vendor
evaluation, and network management

•Marketing. Marketing ES include customer relationship management, mark; analysis, product planning, and market
planning.

•Human resources. Examples of human resources ES are human resources pin- I ning, performance evaluation, staff
scheduling, pension management, and lega | advising.

•Manufacturing. Manufacturing ES include production planning, quality mar. ment, product design, plant site selection, and
equipment maintenance and re:

•Homeland security. Homeland security ES include terrorist threat assessment terrorist finance detection.

•Business process automation. ES have been developed for help desk automation call center management, and regulation
enforcement.

•Health care management. ES have been developed for bioinformatics and other health care management issues
Now that you are familiar with a variety of different ES applications, it is time to look at the internal structure of an ES and
how the goals of the ES are achieved.

ES can be viewed as having two environments: the development environment and the consultation environment (see Figure
12.3). An ES builder uses the development environment to build the components and put knowledge into the knowledge base.
A non-expert uses the consultation environment to obtain expert knowledge and advice. These environments can be separated
when a system is complete.
The three major components that appear in virtually every ES are the knowledge base, inference engine, and user interface. In
general, though, an ES that interacts with the user can contain the following additional components:

•Knowledge acquisition subsystem

•Blackboard (workplace)

•Explanation subsystem (justifier)

•Knowledge refining system


Currently, most ES do not contain the knowledge refinement component. A brief description of each of these components
follows.

KNOWLEDGE ACQUISITION SUBSYSTEM

Knowledge acquisition is the accumulation, transfer, and transformation of problem-solving expertise from experts or
documented knowledge sources to a computer program for constructing or expanding the knowledge base. Potential sources
of knowledge include human experts, textbooks, multimedia documents, databases (public and private), special research
reports, and information available on the Web.
Currently, most organizations have collected a large volume of data, but the organization and management of organizational
knowledge is limited. Knowledge acquisition deals with issues such as making tacit knowledge explicit and integrating
knowledge from multiple sources.

Acquiring knowledge from experts is a complex task that often creates a bottleneck in ES construction. In building large
systems, a knowledge engineer or knowledge elicitation expert needs to interact with one or more human experts in building
t: el knowledge base. Typically, the knowledge engineer helps the expert structure the problem area by interpreting and
integrating human answers to questions, drawing analogies, posing counterexamples, and bringing conceptual difficulties to
light.

KNOWLEDGE BASE

The knowledge base is the foundation of an ES. It contains the relevant knowledge necessary for understanding, formulating,
and solving problems. A typical knowledge base may include two basic elements: (1) facts such as the problem situation and
the theory of the problem area, and (2) special heuristics or rules that direct the use of knowledge that solve specific
problems in a particular domain. (In addition, the inference engine can include general-purpose problem-solving and
decision-making rules.) The heuristics express the informal judgmental knowledge in an application area. Knowledge, not
mere facts, is the primary raw material of ES.

It is important to differentiate the knowledge base of an ES and the knowledge base of an organization. The knowledge
stored in the knowledge base of an ES is of:; J represented in a special format and can be used to solve a particular problem.
Ti J organizational knowledge base, however, contains various kinds of knowledge in different formats and may be stored in
different places. The knowledge base of an ES is only a small subset of an organization's knowledge base.

INFERENCE ENGINE

The "brain" of an ES is the inference engine, also known as the control structure or the rule interpreter (in rule-based ES).This
component is essentially a computer program that provides a methodology for reasoning about information in the knowledge
b a s e and on the blackboard and for formulating conclusions. The inference engine provides directions about how to use the
system's knowledge by developing the agenda that organizes and controls the steps taken to solve problems whenever
consultation takes place.

USER INTERFACE

An ES contains a language processor for friendly, problem-oriented communication between the user and the computer,
known as the user interface. This communication can best be carried out in a natural language. Due to technological
constraints, most exiting systems use the question-and-answer approach to interact with the user. Sometimes it is
supplemented by menus, electronic forms, and graphics.

BLACKBOARD (WORKPLACE)

The blackboard is an area of working memory set aside as a database for the description of a current problem, as specified by
the input data; it is also used for recording intermediate hypotheses and decisions. Three types of decisions can be recorded J
the blackboard: a plan (i.e., how to attack the problem), an agenda (i.e., potential actions awaiting execution), and a solution
(i.e., candidate hypotheses and alternative courses of action that the system has generated thus far).

Consider an example. When your car fails, you can enter the symptoms of the failure into a computer for storage in the
blackboard. As the result of an intermediate hypothesis developed in the blackboard, the computer may then suggest that you
do some additional checks (e.g., see whether your battery is connected properly) and ask you to report the results. This
information is recorded in the blackboard.

EXPLANATION SUBSYSTEM (JUSTIFIER)

The ability to trace responsibility for conclusions to their sources is crucial both in the transfer of expertise and in problem
solving. The explanation subsystem can trace such responsibility and explain the ES behavior by interactively answering
questions such as these:

•Why was a certain question asked by the ES?

•How was a certain conclusion reached?

•Why was a certain alternative rejected?


•What is the plan to reach the solution? For example, what remains to be established before a final diagnosis can be
determined?

In simple ES, the explanation shows the rules that were used to derive the specific recommendations.

KNOWLEDGE-REFINING SYSTEM

Human experts have a knowledge-refining system; that is, they can analyze their own knowledge and its use, learn from it, and
improve on it for future consultations. Similarly, such evaluation is necessary in computerized learning so that a program can
analyze the reasons for its success or failure. This could lead to improvements that result in a more accurate knowledge base
and more effective reasoning.

The critical component for knowledge refinement is the machine-learning component that allows the system to adjust its
activities based on evaluation of past performance. Such a component is not yet mature in commercial ES at the moment but
is being developed in experimental ES at several universities and research institutions.

KNOWLEDGE REPRESENTATION AND ORGANIZATION

Expert knowledge must be represented in a computer-understandable format and organized properly in the knowledge base
of an ES. There are many different ways of representing human knowledge, including production rules, semantic networks,
and logic statements; detailed descriptions of these are provided in Chapter 17, an online chapter. In rule-based systems,
knowledge in the knowledge base is represented in if-then rules that combine the condition and the conclusion for handling a
specific situation. The IF part indicates the condition for the rule to be activated, and the THEN part shows the action or
conclusion if all IF conditions are satisfied.

Let's again consider the example of selecting a notebook computer. Several rules can be defined for choosing between a
Toshiba Satellite A100 and a Dell Latitude XI:

Rule l:

IF the primary task = word processing AND primary usage = travel THEN weight requirement = light

Rule 2:

IF the primary task = word processing

AND primary usage = office

THEN weight requirement = don't care

Rule 3:

IF budget <=2,000

AND budget > 1,000

AND weight requirement = light

THEN NB model = Dell Latitude X I

Rule 4:

IF budget < 1,000

AND weight requirement = don't care THEN NB model = Toshiba Satellite A100

The advantage of using production rules is that this method is easy to understand, and new rules can easily be added to the
knowledge base without affecting existing rules. Uncertainty associated with each rule can be added to enhance its accuracy.
Uncertainty processing is a topic that is discussed in Chapter 13.

A major task of ES development is to acquire knowledge from human experts and then convert it into production rules that
an inference engine can handle. The inference engine chooses applicable rules from the knowledge base, integrates them,
and then reasons to find the conclusion.
THE INFERENCE PROCESS

In complex decisions, expert knowledge often cannot be represented in single rules. Instead, rules may be chained together
dynamically to cover numerous conditions. The process of chaining multiple rules together based on available data is called
inference. The component that conducts inference in an ES is called the inference engine. Two popular approaches for
inferencing are forward chaining and backward chaining. More reasoning strategies, such as case-based reasoning and
advanced techniques, are discussed in Chapters 13 and 14.

Forward Chaining

Forward chaining looks for the IF part of a rule first. When all IF conditions are met, the rule is chosen for deriving the
conclusion. If the conclusion derived from the first state is not final, then it is used as a new fact to match with the IF
condition of other rules to find a more useful conclusion. This process continues until a final conclusion is reached.

Think again about the notebook selection rule base described in this section. In this example, forward chaining would check
two IF conditions associated with Rules 1 and 2: whether the primary task is word processing and whether the primary usage
is in office or travel. If the task is word processing and usage is travel, then Rule 1 applies, and the weight requirement is set
to light. Because the weight requirement is not the final goal, the inference engine further searches the knowledge base and
finds that one IF condition of Rule 3 matches the conclusion from Rule 1. Because Rule 3 has more IF conditions to be
checked, the inference engine triggers Rule 3 and checks whether the price condition is satisfied. If the price condition is also
satisfied, then Rule 3 is activated, and the system recommends Dell Latitude XI to the user. The sequence of activating rules
in the rule base is Rule 1 -» Rule 2 -» Rules 3 or 4 (depending on the result in the previous step).

Backward Chaining

Backward chaining is the reverse of forward chaining. It starts from the conclusion and hypothesizes that the conclusion is
true. The inference engine then identifies the IF conditions necessary for making the conclusion true and locates facts to test
whether the IF conditions are true. If all IF conditions are true, then the rule is chosen, and the conclusion is reached. If some
conditions are false, then the rule is discarded, a n d the next rule is used as the second hypothesis. If there are not adequate
facts to prove that all IF conditions are true or false, the inference engine continues to look for rule whose conclusion would
match the undecided IF condition to move a step further f o r checking the conditions. This chaining process continues until
a set of rules is found I reach a conclusion or to prove unable to reach a conclusion.

Using the notebook selection example, the inference process of backward chaining starts from Rule 3. The inference engine
assumes the Dell Latitude XI to be a product for recommendation and checks the IF conditions of weight requirement and
budget. Because weight requirement is the conclusion of Rules 1 and 2, the engine chains these two rules with Rule 3. The
new IF conditions become primary task and primary usage .The sequence of activating rules in this case is Rule 3 —► Rule
1 —1 Rule 2.

DEVELOPMENT PROCESS OF ES

ES development is a process for eliciting knowledge from experts and then storing the knowledge in the knowledge base.
Because human experts may be unwilling or unable to articulate their knowledge, knowledge acquisition is a critical and
tough task.

A typical process for developing ES includes knowledge acquisition, knowledge representation, selection of development
tools, system prototyping, evaluation, am improvement. Because the inference engine is common to different systems, ES
shell are useful tools that can implement a rule-based system in a very short time. An ES shell is an ES without the
knowledge in the knowledge base. The system runs when the knowledge is stored in the knowledge base.

ES can be classified in several ways. One way is by the general problem areas the; address. For example, diagnosis can be
defined as "inferring system malfunctions from observations." Diagnosis is a generic activity performed in medicine,
organizational studies, computer operations, and so on. A brief description o each category follows:

•Interpretationsystems infer situation descriptions from observations. This category includes surveillance, speech
understanding, image analysis, signal interpretation, and many kinds of intelligence analyses. An interpretation system
explains observed data by assigning them symbolic meanings that describe the situation.

•Prediction systems include weather forecasting; demographic predictions; economic forecasting; traffic predictions; crop
estimates; and military, marketing, and financial forecasting.

1
Problem Area Suitable For Expert System

•Diagnostic systems include medical, electronic, mechanical, and software diagnoses. Diagnostic systems typically relate
observed behavioral irregularities to underlying causes.
•Design systems develop configurations of objects that satisfy the constraints of the design problem. Such problems include
circuit layout, building design, and plant layout. Design systems construct descriptions of objects in various relationships
with one another and verify that these configurations conform to stated constraints.
•Planning systems specialize in planning problems, such as automatic programming. They also deal with short- and long-
term planning in areas such as project management, routing, communications, product development, military applica-
tions, and financial planning.
•Monitoring systems compare observations of system behavior with standards that seem crucial for successful goal
attainment. These crucial features correspond to potential flaws in the plan. There are many computer-aided monitoring
systems for topics ranging from air traffic control to fiscal management tasks.
•Debugging systems rely on planning, design, and prediction capabilities for creating specifications or recommendations to
correct a diagnosed problem.
•Repair systems develop and execute plans to administer a remedy for certain diagnosed problems. Such systems incorporate
debugging, planning, and execution capabilities.
•Instruction systems incorporate diagnosis and debugging subsystems that specifically address the student's needs. Typically,
these systems begin by constructing a hypothetical description of the student's knowledge that interprets her or his
behavior. They then diagnose weaknesses in the student's knowledge and identify appropriate remedies to overcome the
deficiencies. Finally, they plan a tutorial interaction intended to deliver remedial knowledge to the student.
Category Problem Addressed

Interpretation Inferring situation descriptions from observations

Prediction Inferring likely consequences of given situations


Inferring system malfunctions from observations
Diagnosis

Design Configuring objects under constraints

Planning Developing plans to achieve goals

Monitoring Comparing observations to plans and flagging exceptions

Debugging Prescribing remedies for malfunctions

Repair Executing a plan to administer a prescribed remedy

Instruction Diagnosing, debugging, and correcting student


performance

Control Interpreting, predicting, repairing, and monitoring system


behaviors

Вам также может понравиться