III: Decision Support, Expert and Knowledge-based Systems

Question 1

What are the similarities and differences between Decision Support, Expert and Knowledge-Based Systems?

Answer 1

These systems belong in the same category of systems that make decisions from available data. What all these systems have in common is that they support decision-making in a certain setting, handling narrow intelligence.

The general distinguishing factor between these systems is that decision support systems (DSSs) can be simple systems that can support decision-making in any domain without requiring a strict internal modular program structure. While KBs and ESs are designed to be knowledge-based systems.

Question 2

Shortly explain Decision Support, Expert and Knowledge-Based Systems:

Answer 2

Decision support systems - help draw conclusions, can be domain independent and can work with a less structured domain knowledge. DSSs leak creativity, imagination and intuition while being limited by the running computer systems.

Expert systems work within a given domain and work as experts within that domain.

Knowledge-based systems also work within a given domain but are general purpose and the system is modular, thus making it possible to exchange the domain knowledge.

All of them are categorized as intelligent systems.

Question 3

What does it mean when a system is intelligent?

Answer 3

Intelligent systems include almost any system that improves decision-making and problem-solving performance and has a kind of function that works with narrow AI tasks.

Question 4

What does a decision support system do?

Answer 4

A decision support system may only collect and provide data to the users and thereby assist the users’ decision-making, without requiring it to draw any conclusion or give advice. It is still considered a decision support system. One defining feature is the use of an inference mechanism, an organized interpretation of internal facts and knowledge, to draw conclusions or give advice from the data at hand. Another common feature is that they can more or less automatically draw conclusions.

Question 5

In what ways can decision support systems be applied?

Answer 5

Diagnosis of problems and faults, Risk management, Process control och machines and software, Entertainment and education

Question 6

What is Eliza?

Answer 6

ELIZA was intended to act as a doctor and communicated with humans by using basic natural language processing which gave an illusion of understanding.

Question 7

What is DENDRAL?

Answer 7

DENDRAL is a system that analyzes mass spectra data and thereby identifies organic molecules. It could automate the decision-making process. Systems like MYCIN which can identify blood infections and recommend medicine were inspired by DENDRAL. MYCIN was the first knowledge-based system.

Question 8

What categories of Decision support systems are there? Explain them shortly: communication, data-driven, document-driven, knowledge-driven and model-driven

Answer 8

Communication-driven DSSs support conducting meetings by enabling communication between people and facilitating information sharing using an interactive environment. It supports collaborations between activities and group decision tasks.

Data-driven DSSs utilize advanced algorithms and machine learning techniques to process and interpret data, providing decision-makers with valuable information for strategic planning, problem-solving, and performance evaluation. By enabling data-driven decision-making, these systems enhance efficiency, accuracy, and effectiveness in decision-making processes.

Document-driven DSSs employ techniques like natural language processing and text mining to analyze and interpret the textual content, enabling users to gain meaningful insights, identify patterns, and make informed decisions. These systems facilitate information retrieval, knowledge management, and collaborative decision-making by organizing, indexing, and retrieving relevant documents.

Knowledge-driven decision support systems (DSSs) aid decision-making by utilizing expert knowledge and rules. Knowledge-driven DSSs use predefined rules and algorithms to analyze data and generate solutions based on the available knowledge. They are particularly useful in complex domains where expertise and experience are crucial. These systems enable users to model and automate decision-making processes, capture and store knowledge, and apply it consistently to solve problems

Model-driven decision support systems (DSSs) assist decision-making by utilizing mathematical and analytical models. Model-driven DSSs allow users to create and evaluate different scenarios, assess the potential outcomes of decisions, and identify the most optimal solutions based on the defined models. They enable users to understand the implications of their decisions and make informed choices by considering the impact on various factors and constraints.

Question 9

What is the content of decision support systems?

Answer 9

Search engine

Database: contains pre-store facts but can be expanded. Data can come from inputs or external computer system sources.

User Interface

Question 10

What is Business intelligence?

Answer 10

Business intelligence (BI) supports organizations, municipalities or companies to understand trends, businesses and the effects of the environment on their activities. BI is a collective term for skills, techniques, applications, processes and methods. BI processes data by extracting, transforming, managing and analyzing data while also assessing risks and opportunities. It uses different types of methods in order to turn raw data into something of meaning for the business.

Question 11

What are some common business intelligence techniques?

Answer 11

Reporting
Text mining
Text analysis
Data mining
Business performance management
Predictive analytics

Question 12

How does BI process data?

Answer 12

BI processes data by extracting, transforming, managing and analyzing data. Some key features are:

Data can be extracted from multiple data sources representing different business units.

Data is analyzed by techniques
Based on the results of the data analysis situational awareness with a deep understanding of the current situation can be obtained.

Question 13

What is Internet-Of-Things? What is big data?

Answer 13

Internet-of-Things (IoT) is made up of equipment that connects all kinds of physical parts to the Internet. The more IoT devices streaming, the more data sets will grow big data. Big data is an important part for extracting and handling data created by cellphones and web services but also tv, radio and other media.

Question 14

What are the limits of DSSS?

Answer 14

Just like all systems, decision support systems have flaws. DSSs lack creativity, imagination and intuition- Furthermore they are often limited by the computer system in which the DSS is running. Limitations have included interfaces that have not been sophisticated.

Question 15

What are expert systems?

Answer 15

Expert systems are a kind of decision support systems with the difference that decision support systems support decisions while expert systems automatically draw conclusions as an expert. An expert system represents and reasons with the knowledge of some specialists with the aim of solving problems or giving advice.

Question 16

What is MYCIN, Eurisko and Traveller?

Answer 16

They are different types of expert systems. MYCIN is a medical diagnosis tool, Eurisko a heuristic system and Traveller for playing naval war games with a fleet of warships. In all of these systems, there is a knowledge base and a reasoner which is generic, meaning that it can be used in multiple domains by providing a new knowledge base.

Question 17

What is the content of expert systems?

Answer 17

Expert systems are made up of parts needed to automatically draw conclusions. The content has four parts:

A database which contains facts

A user interface which serves as the communication between the user and the system

A domain-specific knowledge base which can be represented as bases, strategies and structures.

Inference mechanism which evaluates the knowledge and executes the content of the knowledge base and the dataset.

Question 18

What is the knowledge base?

Answer 18

The knowledge base is “the brain” of the system. IN the knowledge base, knowledge is represented in a structured way in order to be interpreted by the system. These structures are like composite statements about a domain, but they are also boundaries, constraints and negation (negation meaning logic or failure). The knowledge is built as nested statements, which during execution are combined with other statements and facts.

Question 19

What is knowledge representation?

Answer 19

Knowledge representation (KR) is used to represent knowledge in the knowledge base. Knowledge representation is the structure in which the knowledge is stored in the knowledge base.

Question 20

What are production rules?

Answer 20

They are the most commonly used knowledge representation. The reason is that many find these rules to be easy to create and understand but also easy to interpret during execution. Production rules have the structure of “If condition then conclusion or else another conclusion”.

Question 21

What are subsumed rules? What are circular rules? What are conflicting rules?

Answer 21

Subsumed rules imply that there are several almost identical rules, but one of the rules ahs more details than another rule and they both provide the same answer. This means that the more complex rule should be deleted.

Circular rules are rules that have unfortunately ended up in vicious circles. This means that one rule needs other rules to be satisfied before it can reach a conclusion.

Conflicting rules are rules that give conflicting conclusions which means that one rule provides a conclusion that the other rule will falsify.

Question 22

What is heuristics?

Answer 22

Heuristics, “rule of thumb”, are production rules that leave out information and do not have all parts presented within the rules. That is, not all facts are known and presented in the rule but it is still possible to reach a conclusion. This is common when solving well-known problems that are based on experience.

Question 23

What are metarules?

Answer 23

Metarules, higher-level rules, govern or describe the behavior of other rules within a system. They define the relationships, constraints, or conditions under which other rules should be applied or modified. Metarules provide a framework for organizing and controlling the execution of rules, allowing for greater flexibility and adaptability. Metarules are often used in rule-based systems, expert systems, and artificial intelligence applications to guide the behavior and decision-making processes of rule engines or inference engines.

Question 24

What are frames?

Answer 24

Frames capture the essential characteristics and relationships of an entity, organizing knowledge in a hierarchical manner. They allow for efficient storage, retrieval, and reasoning about complex information by representing objects as collections of attributes and values. Frames provide a structured framework for modeling and understanding the world, facilitating knowledge representation, inference, and problem-solving in various domains.

Question 25

What are semantic networks?

Answer 25

Semantic networks are a graphical representation of knowledge that depict relationships between concepts using nodes and edges. Each node represents a concept, and the edges represent the connections or semantic relationships between them, such as “is-a,” “part-of,” or “related-to.” Semantic networks provide a visual and intuitive way to capture and organize knowledge by representing the meaning and associations between concepts. They enable efficient retrieval and inference by leveraging the interconnected nature of concepts.

Question 26

What is O-A-V?

Answer 26

O-A-V knowledge representations show the relationship between object, attribute and value.

Question 27

What is an ontology?

Answer 27

Ontologies are explicit formal specifications denoting the terms in the domain and relationships among these. Ontologies define a common vocabulary in order to support a mutual understanding of the structure of information among people or software agents and to enable reusing domain knowledge.

Question 28

What are inference mechanism?

Answer 28

Inference mechanisms are processes used in artificial intelligence systems to derive new information or draw logical conclusions based on existing knowledge and rules. These mechanisms utilize logical reasoning, pattern matching, or probabilistic methods to make deductions, predictions, or decisions. They operate by applying logical rules, algorithms, or statistical models to infer additional information from the available data or knowledge.

Question 29

Shortly explain reasoning strategies: deductive, inductive and abductive

Answer 29

Reasoning is the process where new information is derived and combined with facts and the newer information. This is a computer-based way of solving problems. There are several reasoning strategies such as:

Deductive reasoning: if A is a man and all men are mortal then A is mortal

Inductive reasoning: If A is a man and A is mortal, all men named A are moral.

Abductive reasoning: If A experiences a shortness of breath on cold days, and A feels regular shortness of breath when exercising they have asthma.

Case-based reasoning are conclusions based on earlier known and documented cases that correspond to experiences.

Question 30

What is Bayes theorem?

Answer 30

Bayes’ theorem is a fundamental concept in probability theory that describes how to update the probability of an event based on new evidence. It calculates the conditional probability of an event given prior knowledge and the likelihood of the evidence. The theorem states that the updated probability of an event equals the prior probability multiplied by the likelihood of the evidence, divided by the probability of the evidence occurring regardless of the event. Bayes’ theorem has wide applications in fields such as statistics, machine learning, and decision analysis, enabling the incorporation of new information to make more accurate predictions or estimations.

Question 31

What does the Dempster-Shafer theory of evidence say?

Answer 31

The Dempster-Schaffer theory of evidence is a generalization of the Bayesian theory of subjective probability. This theory allows combining evidence from different sources and the Dempster-Schaffer theory arrives at a degree of belief. This degree of belief primarily depends on the number of answers and the subjective probability of each answer.

Question 32

What is a knowledge-based system?

Answer 32

Knowledge-based system is a broad term that refers to many different kinds of systems. These systems utilize data, information and knowledge from different sources in order to generate conclusions. While expert systems refer to the type of task the system is trying to assist in and being able to draw conclusions as a human expert in a complex task.

Question 33

What are the contents of a knowledge-based system?

Answer 33

Database
User interface
Knowledge base
Inference mechanism
Knowledge-based systems have the module for the database in a different place than the rest of the modules.

Question 34

What are shells within an expert system? IS there a well-known shell system?

Answer 34

Expert system shells and knowledge-based system shells include everything except the knowledge in the knowledge base, making the shell ready to be developed. By using shells, the knowledge engineer does not have to build the systems from scratch since developing new systems is mostly about developing the knowledge base.

The E-version of MYCIN (a medical database) is considered a shell system because it works similarly to MYCIN without the domain knowledge.

Question 35

What are the knowledge types in a knowledge-based system? Procedural, declarative, meta-knowledge and heuristics

Answer 35

Procedural is knowledge about how to do something. It is knowledge often learned while carrying out a task, navigating where the task has an order when practicing it.

Declarative describes what is known about an area or a problem, or it can be a map with actions while moving around in the environment and so on. Declarative knowledge is often simple true or false statements in the domain.

Meta-knowledge is knowledge about knowledge and incorporates other instances of existing knowledge and picks the knowledge best suited for solving a problem.

Heuristics is knowledge without having all the data, information or knowledge at hand. Heuristics can be expressed as production rules without some parts.

Question 36

What are cognitive maps?

Answer 36

Cognitive maps are graphs that can present visual images of mental models for decision-making. Cognitive maps are also used for intelligent agents models of environments and pattern recognition in machine learning. Cognitive maps show relationships between concerts and objects using nodes and links.

Question 37

What are inference maps?

Answer 37

Inference networks offer a graphic view of the system’s rules. These rules are represented as nodes and the relationships are links using AND/OR. Using this structure, it is easy to grasp relationships between rules.