Knowledge Engineering (KE) - What and Why

Vocabulary practice flashcards covering definitions, foundational concepts, historical systems, methodologies, and stages of Knowledge Engineering.

Knowledge (Nonaka, 2002)

A justified true belief.

Knowledge (D. Karagiannis, 1998)

Humanized information; more than a static encoding of facts, it includes the ability to use those facts in interacting with the world.

Cognition

The mental act or process by which knowledge is acquired, including perception, intuition, and reasoning.

Logic

The science that investigates the principles governing correct or reliable inference (conclusion).

Certainty

A knowledge concept stating that truths are hard if not impossible to deny.

Evidence

A knowledge concept stating that truths are based on something.

Practicality

A knowledge concept stating that truths have to actually work in the real world.

Broad Agreement

A knowledge concept stating that truth is based on the agreement of lots of people.

Expert Knowledge

Information woven inside the mind of an expert for accurately and quickly solving complex problems.

Facts of the Domain

The widely shared knowledge written in textbooks and journals of a field, constituting the type of material a professor would lecture about in class.

Heuristic Knowledge

Knowledge that constitutes the rules of expertise, rules of good practice, judgmental rules of the field, and rules of plausible reasoning.

Knowledge Base

A set of representations of facts about the world, encompassing knowledge of the problem domain and knowledge of how to solve the problem.

DENDRAL

An early chemical-analysis expert system developed beginning in 1965 by Edward Feigenbaum and Joshua Lederberg at Stanford University whose performance rivaled human chemist experts.

MYCIN

A medical diagnosis expert system developed for Edward Shortliffe's PhD dissertation (1972-1974) at Stanford that used about 500 production rules to handle diagnoses under uncertainty.

CADUCEUS

A medical expert system completed in the mid-1980s by Harry Pople at the University of Pittsburgh that could diagnose up to 1000 different diseases and was described as the most knowledge-intensive expert system in existence.

SHYSTER

A legal expert system developed by James Popple (1993) at the Australian National University that uses a case-based approach to model how lawyers argue with cases.

Modeling Relation (Robert Rosen)

The creative, unentailed comparison of inferential structures between a natural system and a formal system via encoding and decoding arrows.

Knowledge Engineering

The process of building, maintaining, and developing intelligent knowledge-based systems such as Expert Systems and Decision Support Systems.

Transfer View

The traditional view of knowledge engineering assuming concrete human knowledge already exists to solve a problem and must be transferred and implemented directly into the knowledge base.

Modeling View

The alternative view of knowledge engineering treating problem solving as a dynamic, cyclic, continuous process dependent on acquired knowledge and system interpretations.

Knowledge Acquisition

The scientific issue of obtaining knowledge automatically or semi-automatically to facilitate the transfer of expert knowledge into machine data structures.

Knowledge Representation

The method of structuring domain knowledge as data structures in computer memory so they can be conveniently accessed for problem solving.

The Representation Principle

The principle stating that once a problem is described using an appropriate representation, the problem is almost solved.

Knowledge Utilization

The scientific issue concerning how knowledge is used in problem solving, primarily dealing with the design and operation of the inference engine.

Phase 1: Problem Assessment

The knowledge engineering phase dedicated to determining problem characteristics, identifying participants, specifying objectives, and determining needed resources.

Diagnosis (Problem Type)

Inferring malfunctions of an object from its behaviour and recommending solutions.

Selection (Problem Type)

Recommending the best option from a list of possible alternatives.

Prediction (Problem Type)

Predicting the future behaviour of an object from its behaviour in the past.

Classification (Problem Type)

Assigning an object to one of the defined classes.

Clustering (Problem Type)

Dividing a heterogeneous group of objects into homogeneous subgroups.

Optimisation (Problem Type)

Improving the quality of solutions until an optimal one is found.

Control (Problem Type)

Governing the behaviour of an object to meet specified requirements in real-time.

Phase 2: Data and Knowledge Acquisition

The knowledge engineering phase focused on collecting and analyzing data, making design concepts explicit, and resolving incompatible, inconsistent, or missing data.

Phase 3: Development of a Prototype System

The knowledge engineering phase involving tool selection, knowledge representation, and prototype implementation and testing.

Prototype System

A small version of the final system built to verify understanding of the problem, ensure adequacy of tools and representations, and engage experts and skeptics.

Test Case

A problem successfully solved in the past for which input data and the resulting output solution are already known.

Phase 4: Development of a Complete System

The knowledge engineering phase focused on preparing full-scale system design, collecting additional knowledge, building the user interface, and implementing the full system.

Phase 5: Evaluation and Revision of the System

The knowledge engineering phase in which the system is tested against performance criteria agreed upon during prototyping and revised to ensure user satisfaction.

Phase 6: Integration and Maintenance

The knowledge engineering phase where arrangements for technology transfer are made and an ongoing maintenance program is established.

Expert System

A knowledge-based program containing expert knowledge that provides expertise for solving problems in a defined application area in the manner of human experts.

Fuzzy Expert System

A rule-based expert system utilizing fuzzy rules and inference to process common sense knowledge that is subjective, ambiguous, vague, or contradictory.

Artificial Neural Network

A biologically inspired connectionist computational model consisting of processing elements (neurons), interconnecting links, and training and recall algorithms.

Hybrid System

An intelligent system integrating multiple paradigms (such as rule-based systems, fuzzy systems, neural networks, and genetic algorithms) to combine their distinct advantages.

Statistical Methods (in KE)

Methods applicable when statistically representable data are available and the underlying type of the goal function is known.

Symbolic AI Systems

Rule-based systems using propositional logic, predicate logic, or production systems, suited for problems with well-defined, rigid rules where adaptation is minimal or difficult.

Genetic Algorithms (in KE)

Search and optimization methods requiring neither data sets nor heuristic rules, but only a simple selection criterion, proving efficient when little is initially known.