Harvard-MIT Division of Health Sciences and Technology
HST.952: Computing for Biomedical Scientists
Data and Knowledge
Representation
Lecture 6
Last Time We Talked About
z Major
KR schemes
z Semantic
Network
z Frame-based Representation
z Production Rules
Today We Will Talk About
z Process
z Finite
state machine
z Flow chart
z Petri Net
Process
z Object:
z E.g.
continuant
diagnosis, medication, data repository
z Process:
z E.g.
occurrent
diagnose, treat, retrieve
Describing Process
z Predicate
z treat
logic
(Dr. Jones, Ms. List)
z Frame-based
system
z Patient
Frame has a slot “primary care” or
“attending physician”
z There
are more about “treat” to specify
z When,
where, how, why, what result
Types of Process
Process
Continuous Process
Discrete Process
Event
Cessation
Initiation
Continuation
State
Example
z Continuous
Process
z Aging
z Discrete
z Out
Process
patient visit (make appointment, check in
with nurse, see a doctor, have test/receive
medication)
Basic Distinctions
z Discrete
or continuous
z Linear or branching
z Independent or ramified
z Immediate or delayed
z Sequential or concurrent
z Predictable or surprising
Basic Distinctions
z Normal
or equinomral
z Flat or hierarchical
z Timeless or time-bound
z Forgetful or memory-bound
Process, Procedure and History
z Sequence
of events and state
z Process: one event or state is current
z Procedure: abstract (pattern or script of
processes)
z History: record of a past process
Finite State Machine (FSM)
z Discrete
process (continuous process can
be simulated with fine time steps)
z State transition diagram
z Formal Definition:
zA
finite set of states: Q
z A finite set of inputs: I
z A transition function F(Q, I) -> Q, F can be a
partial Function
FSM
z Basic
components: state (circle), transition
(arrow), input (label on the arrow)
z End state can be marked with double
circles
FSM
normal blood sugar in test 1
No DB
Suspect DB
High blood sugar in test 1
Highly suspect
DB
normal blood sugar in test 2
High blood sugar in test 2
Confirmed DB
Flow Chart
z Event
(box) and decision (diamond)
z Arrow (transition) and label on the arrow
(condition for decision)
z Start and end can be specified with box
Petri Net
zA
finite set of places (circle)
z A finite set of transactions (line)
z A finite set of arrows connecting either
places to transactions or transactions to
places
Petri Net
z Marking:
assign a non-negative integer to
each place. (dot/token)
z Firing: a transaction take place after
enabled
z Firing sequence: the sequence of
transaction firings for a given PN with a
given initial marking
Petri Net
z Conflict
z Starvation
z Deadlock
Petri Net
z Assign
values to tokens
z Define functions for transaction
z Specify scheduling policies
z Times Petri Net
Exercise
Overweight patients should be instructed
either excise regularly or keep a healthy
diet for 3 months. When one fails, try both
for 3 months. When both fail, a patient
should be given medication x for 2
months.
Reading
z Sowa
Chap. 3