|
Additions
to this glossary are always welcome so please email
us with suggestions.
Thanks go to
at
Herriot-Watt University, UK for providing the foundations of this
glossary.
2nd Revision 5th August 1998 by the MONET Communications
& Infrastructure Committee
Glossary
A - B
- C - D
- E - F
- G - H
- I - J
- K - L
- M - N
- O - P
- Q - R
- S - T
- U - V
- W - X
- Y - Z
- Abduction
- Reasoning
to the (best) explanation.
Also reasoning from effects to causes. In diagnosis it is ofetn
used in conjunction with through fault
models.
- Abstraction
- The degree
of disassociation from any specific instance.
- Accuracy
-
- The continuous
version of correctness.
- An emergent
property of the system, that may only be determined with regard
to a pre-defined precision & uncertainty. Absolute
accuracy is in terms of the precision of the gold standard,
while relative accuracy is defined with the quantity
space of the model and is relative to the abstraction of gold
standard behaviour in that space.
- Association
-
- A correlation
between a state and a conclusion, a condition-action pair
being a rule. Often empirically
derived.
- A correlation
between two objects.
- ATMS
- Assumption-Based
Truth Maintenance System. A misnomer. Actually a consistency maintenance
system.
- Backward
chaining
- Rreasoning
from consequents to premises, that is goal-directed reasoning.
cf forward chaining
- Behaviour
-
- Response
of a system to its environment (As opp. to cognitive) [ITS].
- To act,
function, or react in a particular way.
- Result
of an execution of knowledge.
- Behavioural
choice
- The selection
of the behaviour required.
- (Fault)
Candidate Proposer
-
- Diagnostic
device for proposing possible faults.
- Diagnostic
device for generating models representing the possible fault
hypotheses.
- (Fault)
Candidate Set
- The set of
possible diagnoses of fault hypotheses.
- Causal
- Answers to
questions of the form 'wherefore...?' or, more vulgarly, 'how
come...?' involve causality. cf teleological. Of or related to
something that occasions or effects a result.
- Causal
explanation
- Explains
in terms of a cause preceding or simultaneous to the fact being
explained.
- Causal
propagation
- See qualitative
analysis.
- Causally
ordered
- The set of
constraints which may be applied in order.
- Cognitive
-
- Mental,
as opposed to physical (behavioural).
- Of, or
relating to, unmeasurable,
but observable, mental
processes.
- Cognitive
process
-
- Component
- A part of
a system.
- Conflict
set
-
- The complete
set of assumptions underlying a detected discrepancy.
- A set
of components that
cannot be all functioning correctly (at least one of them
must be faulty
- Constraint
- A relation
between a set of variables. For example an equation or an inequality.
- Correctness
- Boolean.
A behavioural model is correct if its predicted behaviour is consistent
with the facts (i.e., the observations, or reference behaviour)
in the domain.
- Corroboration
- Exhoneration
of the components that
exhibit normal behavior or that are related to observations
corresponding to normal (predicted) behavior.
- Crisp
-
- curly,
wavy; also : having close stiff or wiry curls or waves
- having
the surface roughened into small folds or curling wrinkles
2a: easily crumbled : brittle of pastry
- being
firm and fresh {~lettuce}
- being
sharp, clean-cut, and clear {a ~illustration}
- noticeably
neat.
-
- (fault)
Deduction
- Reasoning
the general to the specific.
- (Fault)
Detection
- The process
by which behavioural deviation is flagged.
- Data
Acquisition
- Getting observations.
- Data
Interpretation
- Pre-processing
observations for reasoning.
- Deduction
- Reasoning
from the general to the specific.
- Dependency-Based
Diagnosis
- Diagnostic
reasoning between predictions and supporting assumptions. (cf
abduction.)
- Discrepancy
-
- An inconsistency
between an observed & predicted value.
- An inconsistency
between prediction & observation.
- Discrepancy
preprocessor
- A mechanism
for producing a composite 'measure of mismatch' from the stream
of raw discrepancies, used by the candidate
proposer.
- Domain
- An area about
which knowledge exists.
- Domain
Knowledge
- Expert knowledge
of a specific subject. Not mathsy.
- Dynamic
- Of, or related
to, systems whose future state
is consequent upon their past & present states; that is, relating
to systems who have a history. Contains differential or integral
expressions. Not static, but may be time-varying.
Dependent
upon box-drawing: if room temperature is taken as an exogenous
factor, for example, then it is time-varying; but
if the weather is included in the model, then it is a dynamic
factor. cf static.
-
- Exoneration
Set
- Set of hypotheses
which appear to be consistent with observations.
- Explanation
- What it says.
Divided into theoretic, justification, description, and instruction
subtypes.
- (Descriptive)
Explanation
- Correlative;
considers goals & choice.
- (Instructive)
Explanation
- Temporal;
considers actions & sequences.
- (Justification)
Explanation
- Deductive;
considers causality, goals, function & choice.
- (Theoretic)
Explanation
- Deductive;
considers causal and scientific laws.
FAULT
-
- (Fault)
Detection
- The decision
that the monitored system is exhibiting faulty behaviour.
- Fault
Identification
-
- Identifying
the new relationships between values in the system which have
changed because of the fault.
- Reasoning
about the type or extent of a given fault.
- Determining
the specific fault of a component
- (Fault)
Explanation
- What it says.
How the fault resulted in the observed faulty behaviour.
- (Fault)
Isolation
- what it says.
Location of fault to part of the monitored system.
- Fault
Model
- Model
of a faulty system.
- Fidelity
- The emergent
model property reflecting the wellposedness of a selected knowledge
in achieving a specific behaviour.
- Forward
Chaining
- reasoning
from premises to consequents; data-directed reasoning. cf backward
chaining.
- Fundamental
Choice
- The basic
selection of model properties, in terms of ontological,
representational
and behavioural
choices.
- Fuzzy
Number
- A compact
convex fuzzy set.
- Fuzzy
Set
- A set over
some domain in which each of the elements has a value between
0 and 1.
- GDE
- General Diagnostic
Engine
- Generality
-
- The degree
of applicability outwith the current application.
- The degree
of value-independence.
- Granularity
- The size
of the divisions in the variable
space.
cf precision.
-
- Identification
-
- Reasoning
about the previous states
of a system
from its current state.
- See Fault
Identification.
- Induction
- Reasoning
from the specific to the general.
- Interpretation
- Translating
raw data (gauge readings, etc.) into symbolic structures.
- Interval
Space
-
- Justification
- Dependency
between at atom and those that have been used for inferring it.
Used in the atms
.
- Knowledge
Classes
- Knowledge
can be classified as one of three, increasingly general, classes:
procedures, associations and principles.
- Landmark
- A distinctive
value for a variable
that
separates values that corresponds to different qualitative
abstractions.
- Landmark-based
domain
- A qualitative
space
(domain) of values for a variable
based
on the segmentation of its quantitative
space
with landmark
values
- Learning
Style
- Not all people
learn most effectively the same way. Using the three knowledge
classes, we can see that the trainee might best learn by induction,
deduction or rote; each these is a style.
- Level
-
-
- Measurable
- A part of
the system
which may be measured directly.
- Measurement
- The result
of measuring a variable.
- Membership
Function
- The weights
by which various items are associated with a set. cf fuzzy
set.
- Model
-
- An executable
description of a system
or body of knowledge.
- The execution
of such a description; that is, description + the inference
engine = model.
- Model-based
Reasoning
- The symbolic
processing of an explicit representation of the internal workings
of a system in order to predict, simulate and explain the resultant
behaviour of the system from the structure, causality, functional
and behaviour of its components.
- Modelling
Primitives
- The set of
basic descriptors used in model construction. Legal operations
within the chosen ontology.
- Network
of Excellence
- This describes
an institution which organises and facilitates communication between
other organisations who are interested in a particular field of
research.
-
- No-function-in-structure
Principle
- The behavior
of a component
must
be described independently of the specific function that the component
can have in different devices and of the specific function that
different instances of the component can have in a device.
-
- Observable
- A part of
the system that may be calculated indirectly from measurables.
- Observation
- The instantiation
of an observable.
- Observer
- A mechanism
for calculating the value of non-measurable variables
from past or present variable values.
- Ontology
-
- The reflection
of the commitment of models over the primitive elements of
the application domain.
- Reality:
what is being modelled.
- Ontological
Choice
- The selection
of knowledge. cf representational
choice
and behavioural
choice.
- Order
of Magnitude
- A qualitative
characterization
of the values of a variable
.
Can be either absolute or relative (to other variables)
- Orientation
- The spectrum
from implicit to explicit knowledge coding.
- Parameter
- A non-dynamic
entity.
- Parameter
Space
-
- Perspicuity
- The
understandability of models
in terms of clarity and user-friendliness.
- Possibilistic
-
- Precision
- The degree
of detail in the value of a variable
or model.
cf Granularity.
- Predictor
- A device
for predicting observations. cf candidate
proposer
- Principle
-
- A highly
general,
ideally fundamental, and often equation-based,item of knowledge.
Inefficient in use.
- A general
rule which has been scientifically,as opposed to empirically,
derived. Inefficient in use.
- Probabilistic
- Of, relating
to, or based on probability.
- Probability
- A mathematical
basis for prediction that for an exhaustive set of outcomes is
the ratio of the outcomes that would produce a given event to
the total number of possible outcomes.
- Probe
- The action
of measuring a variable.
- Procedure
- An ordered
set of actions, possibly composed of subprocedures executed without
conscious thought. Highly efficient and very specific.
- Qualitative
- Of or relating
to a
non-numerical description.
- Qualitative
Analysis
- The application
of model constraints to create the set of consistent qualitative
states. cf transaction
analysis.
- Qualitative
Reasoning
- Of or relating
to a non-numerical description of a system, preserving all its
important behavioural properties and distinctions. Qualitative
Models aim to capture the fundamental aspects of a system or mechanism,
while suppressing much of the detail. Methods such as abstraction
and approximation are often used to build models based on qualitative
rather than numerical aspects of a system.
- Quantitative
- Of or relating
to a, possibly fuzzy,
numerical description.
- Quantity
Space
-
- A value
domain (in the mathematical sense) consisting of a finite
number of qualitative values.
- The set
of quantities used to describe the values of model variables.
- Remediation
- Alleviating
a fault.
- Representational
Choice
- The selection
of a particular encoding of knowledge.
- Resolution
- The number
of components which comprise the system.
- Rule
- A subtype
of association;
a condition-action pair.
- Scope
- The part
of the domain knowledge, or system, represented. Hence defines
the boundary between the model and its environment.
- Semi-Quantitative
- A hybrid
of quantitative & qualitative approaches, intended to prune
spurious behaviours by means of more precise system characterisation.
- Sensor
- A mechanism
for measuring measurables.
- Signs
- A simple
qualitative
domain
for a variable
based
on the sign of the value that the variable assumes
- Simulator
- An enviroment
for the execution of a model.
- Simulation
- The execution
of a model.
- Source
- The spectrum
of origin of domain knowledge, from empirical to theoretical...
- State
- The smallest
set of information required, along with model
& parameters,
to uniquely describe everything about a system.
- State
Space
- The set of
possible states
in a system.
- State
Variable Observer
- See observer.
- Static
- Non-dynamic;
ie, containing only algebraic relations.
- System
- An interacting,
or interdependent, group of entities.
- Teleological
- Exhibiting
or relating to design or purpose.
- Telelogical
Explanation
- Explain the
fact under consideration as being causally related to a goal in
the future.
- Test
- a specific
condition that can be applied to a system
for
making measurements
- Time-Varying
- Of or relating
to change over time which can be described algebraically OR with
calculus. Thus either static
or dynamic.
- Training
Objective
- A milestone
in the development of expertise. Coagulate into a curriculum.
- Trainee
Profile
- A relatively
static description of the trainee, encompassing their preferences
and strengths, used in selection of the training strategy.
- Training
Plan
- Consisting
of training
units,
a training plan is the result of hierarchical decomposition of
a training
objective,
tailored according to the learning preferences and prior knowledge
of the trainee. A plan is the means by which the trainee satisfies
the goal.
- Training
Unit
- A domain-specific
instantiation of a training
unit class.
An atomic 'chunk' of training interaction, which groups services
provided by the domain modules and delivers the training dialogue
via the trainee interface. The basic units that are organised
into training plans. cf Training
Unit Classes.
- Training
Unit Class
- Also known
as training unit types, each of these is an domain independent,
abstract definition of one kind of training interaction. The four
main classes are Presentation, Demonstration, Practice, &
Test.
- Transaction
Analysis
- The postulation
of new qualitative values for model variables by evolving them
over time. cf qualitative
analysis.
- Uncertainty
- A measure
of the confidence in the value of an observable
or a prediction. Usually described in fuzzy or probabilistic
terms.
- User
Model
- An executable
representation of the trainee.
- Variable
- A entity
which may assume any one of a set of values.
- Variable
Space
- The set of
possible values which may be taken by variables.
|
