Knowledge Engineering (IT4362) Quang Nhat NGUYEN - - PowerPoint PPT Presentation

Knowledge Engineering (IT4362)

Hanoi University of Science and Technology School of Information and Communication Technology Academic year 2020-2021

Quang Nhat NGUYEN

(quang.nguyennhat@hust.edu.vn)

Content

◼ Introduction

• Knowledge engineering
• Knowledge-based systems
• Social impact

◼ First-order logic ◼ Knowledge representation ◼ Logic programming ◼ Expert systems ◼ Uncertain reasoning ◼ Knowledge discovery by Machine learning ◼ Knowledge discovery by Data mining

2

Knowledge Engineering

Definition of Data

◼ Data (the plural of datum) are just raw facts (Long and

Long, 1998)

◼ Data . . . are streams of raw facts representing events . . .

before they have been arranged into a form that people can understand and use (Laudon and Laudon, 1998)

◼ Data is comprised of facts (Hayes, 1992) ◼ Recorded symbols (McNurlin and Sprague, 1998)

→Data is often defined as facts or symbols

3

Knowledge Engineering

Definition of Information

◼ That property of data which represents and measures effects

• f processing them (Hayes, 1992)

◼ Data that have been shaped into a form that is meaningful and

useful to human beings (Laudon and Laudon, 1998)

◼ Data that have been collected and processed into a

meaningful form. Simply, information is the meaning we give to accumulated facts (data) (Long and Long, 1998)

◼ Data in context (McNurlin and Sprague, 1998)

→Information is often defined as data processed or transformed into a form or structure suitable for use by human beings →Information comes after (does not appear before) data

4

Knowledge Engineering

Definition of Knowledge

◼ The result of the understanding of information (Hayes,

1992)

◼ The result of internalizing information (Hayes, 1992) ◼ Collected information about an area of concern (Senn,

1990)

◼ Information with direction or intent – it facilitates a

decision or an action (Zachman, 1987) →Knowledge is often defined as understanding of information

5

Knowledge Engineering

Pyramid of Data/Information/Knowledge

(Adapted from “Knowledge Engineering course (CM3016), by K. Hui 2008-2009”)

6

Knowledge Engineering

Meta- Knowledge Knowledge Information Data

Large volume. Low

• value. Usually no

meaning/ context Lower volume. Higher

• value. With context and

associated meanings Understanding of a

• domain. Can be applied to

solve problems Knowledge on knowledge (e.g., how/when to apply) Management information systems Knowledge- based systems Databases, transaction systems

Example of Data/Information/Knowledge

◼ Data

❑ The temperature outside is 5 degree Celsius

◼ Information

❑ It is cold outside

◼ Knowledge

❑ If it is cold outside then you should wear a warm coat

→The perceived value of data increases as it is transferred into knowledge →Knowledge enables useful decisions to be made

7

Knowledge Engineering

Definition of Knowledge Engineering

◼ Knowledge Engineering (KE)

❑ An engineering discipline ❑ To integrate knowledge into computer systems (in order to solve

complex problems normally requiring a high level of human expertise)

◼ KE enables computer systems

❑ To build knowledge bases ❑ To maintain knowledge bases ❑ To exploit knowledge bases to provide solutions to real-world

problems

8

Knowledge Engineering

Main activities of KE

◼ Knowledge representation

❑ To represent (encode) knowledge in the knowledge base

◼ Knowledge acquisition

❑ To obtain knowledge from various sources (e.g., human experts,

computer sources of data, books, etc.)

◼ Knowledge validation

❑ Knowledge is checked using test cases for adequate quality

◼ Inferring (reasoning)

❑ To form inferences in the knowledge so that the system can

make a decision or provide advice to the user

◼ Explanation and justification

❑ To explain how a conclusion was reached using knowledge in the

knowledge base

9

Knowledge Engineering

Knowledge-based systems

◼ Knowledge-based systems (KBSs) are those systems

that maintain and exploit knowledge to solve real-world (often complex) problems

◼ KBS = Knowledge + Inference (Reasoning) ◼ Knowledge in a KBS

❑ Specific to a domain (application area) ❑ Represented and stored in the system’s knowledge base

◼ Inference (reasoning) in a KBS

❑ Apply domain-specific knowledge to search for a solution (if any) ❑ Simulate a logical reasoning process ◼ E.g., Simulate the reasoning process of experts in the given domain 10

Knowledge Engineering

Main features of KBSs

◼ The separation of the knowledge from how it is used ◼ The use of specific-domain knowledge ◼ The heuristic (experience) rather than algorithmic nature

• f the knowledge employed

❑ Solve problems by heuristic methods, not just only by algorithms

◼ Simulate human reasoning about a problem domain

❑ Perform reasoning over representations of knowledge 11

Knowledge Engineering

Architecture of a KBS (1)

12

Knowledge Engineering

Inference engine Knowledge-based system User interface Knowledge base Working memory User

Architecture of a KBS (2)

◼ Knowledge Base (KB)

❑ Representation of knowledge ❑ Contains expertise (domain-specific knowledge), such as

relationships/association between objects/concepts, problem- solving strategies, etc.

◼ Inference Engine

❑ Use of knowledge ❑ Applies knowledge in the KB in order to find a solution to the

current problem

13

Knowledge Engineering

Architecture of a KBS (3)

◼ Working Memory

❑ To temporarily store status of the current problem solving session ◼ facts about the current situation ◼ any (intermediate) conclusions drawn ◼ hypothesis (goals)

◼ User Interface

❑ Allows the KBS to interact with the user ❑ User provides facts (i.e., describes the problem) and queries the

KBS to get a solution

❑ The KBS returns solution(s) 14

Knowledge Engineering

Main steps of development of a KBS

◼ Problem identification and analysis

❑ Elicit requirements (e.g., organizational needs and constraints)

◼ Knowledge engineering

❑ Represent knowledge, acquire knowledge, transfer acquired

knowledge to the knowledge base, construct inference engine

◼ System modeling and design ◼ System implementation

❑ Develop system prototype

◼ System evaluation

❑ Test the knowledge, test the reasoning, and test the system

◼ System maintenance

15

Knowledge Engineering

Knowledge acquisition

◼ To acquire (elicit) knowledge from sources of expertise,

and transfer it to the knowledge base

◼ Sources of expertise: human experts, books, magazines,

databases, the Internet, etc.

◼ Methods of knowledge acquisition

❑ Manually: interviewing, or observing and tracking the reasoning

process

❑ Automatically: using computer programs to discover knowledge ❑ Semi-automatically: interviewing human experts, with the aid

(support) of computer programs/tools

16

Knowledge Engineering

Knowledge representation

◼ A number of knowledge representation methods

❑ Production rules, Frames, Semantic networks, Ontology,

Probabilistic models, etc.

◼ Completeness

❑ Support the acquisition of all aspects of knowledge

◼ Conciseness

❑ Efficient acquisition, Easy storage and access

◼ Computational efficiency ◼ Transparency

❑ Enable understanding of the system’s behaviour and conclusions 17

Knowledge Engineering

Typical types of KBSs

◼ Expert systems

❑ To mimic the decision-making process of human experts in a

specific domain

◼ Machine learning-based systems

❑ To learn from experience (i.e., examples) to discover knowledge ❑ E.g., Neural networks, case-based reasoning, decision tree, etc.

◼ Data mining-based systems

❑ To identify relationships in data (in large datasets) ❑ E.g., To identify products/services often purchased at the same

time

◼ Intelligent agent-based systems

❑ To learn and make increasingly complex decisions on behalf of

their users, in order to reach (achieve) their predefined goal(s)

18

Knowledge Engineering

Typical tasks of KBSs (1)

◼ Decision making support (e.g., decisions of product

selection)

◼ Interpretation of data (e.g., sonar signals, vocal signals) ◼ Diagnosis of malfunctions (e.g., diseases, machine fails) ◼ Structural analysis of complex objects (e.g., chemical

compounds, DNA sequences)

◼ Configuration of complex objects (e.g., computer

systems)

◼ Recognition of objects (e.g., faces, hand-written words)

19

Knowledge Engineering

Typical tasks of KBSs (2)

◼ Classification of concepts (e.g., animals, web pages’

categories)

◼ Prediction of consequences of situations (e.g., storm

forecasting)

◼ Scheduling activities (e.g., schedule of courses) ◼ Planning sequence of actions (e.g., robot’s motion and

actions)

◼ Natural language understanding (e.g., language

translation of a text)

20

Knowledge Engineering

Advantages of KBSs

◼ Distribution and availability of knowledge (expertise)

❑ KBSs enables knowledge to be exploited in any time, and in

anywhere

◼ Consistent results

❑ Given a problem, results returned by different persons (or the

same person at different moments) may be inconsistent

◼ Retaining of knowledge

❑ To store obtained knowledge for reuse in the future

◼ Able to solve problems with incomplete/uncertain data

and knowledge

◼ Able to explain the system’s produced solutions

21

Knowledge Engineering

Disadvantages of KBSs

◼ Correctness of the produced results (solutions) depends

strongly on the knowledge owned by the system

◼ Limited knowledge

❑ Not all knowledge can be represented and captured ❑ The system does not know the limitations of its owned knowledge

◼ Lack of ‘commonsense knowledge’

❑ Knowledge about the same application domain of two KBSs may

be different (even contain some conflicts)

◼ Elicitation and maintenance of knowledge is (very) difficult ◼ Often need (but also often difficult) to manage the degradation

• f knowledge over time

❑ True knowledge at a previous time may be less (or not) correct at

a latter time

22

Knowledge Engineering

Examples of practical KBSs (1)

◼ MYCIN

❑ Work as a consultant for physicians ❑ Diagnose certain infectious diseases (e.g., infectious blood) ❑ Prescribe anti-microbial therapy ❑ Can explain its reasoning in detail ❑ In a controlled test, its performance equaled that of specialists

◼ DENDRAL

❑ Support organic chemists in identifying unknown organic

molecules, and in determining the molecular structure of soil

❑ Uses the molecular formula, the spectrographic data, and the

encoded heuristic knowledge of organic chemists and geneticists

23

Knowledge Engineering

Examples of practical KBSs (2)

◼ PROSPECTOR

❑ For mineral exploration ❑ Use a combined structure of rules and a semantic network ❑ Help exploration geologists assess suspected mineral deposits

◼ XCON

❑ Assist customers in the ordering of DEC's VAX computer systems

– by automatically selecting the computer system components based on the customer's requirements

❑ Before XCON, when ordering a VAX from DEC, every cable,

connection, and bit of software had to be ordered separately

❑ Customers would find that they had hardware without the correct

cables, printers without the correct drivers, a processor without the correct language chip, etc.

24

Knowledge Engineering

Examples of practical KBSs (3)

◼ MEDEX

❑ Predict the onset, continuation and cessation of specific gale-

force wind events (wind speeds greater than 17 meters per second) throughout various regions within the Mediterranean Sea

❑ Uses expert system methods to represent the expertise of a

meteorologist/forecaster with 25 years of experience in the Mediterranean

❑ Use fuzzy set methods to deal with the uncertainty and

imprecision inherent in the expression of this type of knowledge

◼ NAVEX

❑ Monitor radar data ❑ Estimate the velocity and position of the space shuttle 25

Knowledge Engineering

Examples of practical KBSs (4)

◼ CROP ADVISOR

❑ Advise cereal grain farmers on appropriate fertilizers and

pesticides for their farms

❑ Given relevant data, the system produces various financial return

projections for different application rates of different chemicals

❑ The system uses statistical reasoning to come to these

conclusions

◼ OPTIMUM-AIV

❑ A planner, used by the European Space Agency, to help in the

assembly, integration, and verification of spacecraft

❑ Generate plans and monitor their execution ❑ Can reason about complex conditions, time, and resources (such

as budget constraints)

26

Knowledge Engineering

Examples of practical KBSs (5)

◼ FraudWatch (Barclay Bank)

❑ To detect frauds in use of credit cards

◼ KLM Airlines

❑ Help customers construct their flight schedule

◼ Hitachi

❑ Process scheduling in chemical plants

◼ Toshiba

❑ Diagnose faults (and restore operation) of an electric power system

◼ NEC

❑ COSMOS/AI – a crew scheduling system for Japan Airlines

◼ Mitsubishi Electric

❑ A knowledge-based system for elevator group control 27

Knowledge Engineering

References

◼ R. Hayes. The Measurement of Information. In Vakkari, P. and Cronin, B.

(editors): Conceptions of Library and Information Science, pp. 97–108. Taylor Graham, 1992.

◼ K. C. Laudon and J. P. Laudon. Management Information Systems: New

Approaches to Organisation and Technology (5th edition). Prentice-Hall, 1998.

◼ L. Long and N. Long. Computers (5th edition). Prentice-Hall, 1998. ◼ B. McNurlin and R. H. Sprague. Information Systems Management in

Practice (4th edition). Prentice-Hall, 1998.

◼ J. A. Senn. Information Systems in Management. Wadsworth Publishing,

1990.

◼ J. Zachman. A Framework for Information Systems Architecture. IBM

Systems Journal, 26(3): 276–292, 1987.

28

Knowledge Engineering