We need to represent states in a way that makes them persistent, and yet can influence
other actions and states - they have to be linkable objects.
We probably need to differentiate more clearly between actions and
the states they alter, and to have a state relation, so “opening the door”
changes the state relation on the door, which changes a state. The state can be anywhere
in a spectrum, so it can be “open”, "partly open”, “closed”,
all handled by the same state relation. States can also be
related to specific objects, such as
Fred alerted John to the danger.
Now
John is aware of the danger.
John has had a persistent change in state (which may not
persist – John can forget it) – exercised through a relation which the action
created. We could do this, or we could use a SYNONYMMAP,
which can be navigated through during Semantic Searching. That is, if Fred alerted John to
the danger, then John is aware of the danger. It adds a refinement to the SYNONYMMAP
– it has to handle timings as well, but simplifies the network construction.
ToBeAware may have its own
modifiers – John may only be dimly aware of the danger.
Starting with opening doors, but remembering we may need
specific objects:
- The Door Is Open
- The door is open.
- The door is slightly open.
- The door is closed.
- The door is open to newcomers, closed to repeat visitors.
- The contract is open to misinterpretation.
In case 4, it depends who you are as to what state you encounter,
making propagation of the state messy.
In case 5, “is open to misinterpretation” is equivalent to “is able to be
misinterpreted” – we need to be clear on what are openable
things.
For openable things – doors,
meetings – we could operate directly on the state
The state attribute is an invocation of OpenClosed,
which is a spectrum from Open to Closed, making the state bounded and transmissible to
another similar object – it can’t lie outside the spectrum.
This statement can be used to assert the door is open, or query the
state of the door. The state is radiated, and no error is generated if we hit another
relation saying it is closed, but with a different time (the new state may terminate the
time of the other relation). If Fred says “It is open” and John says “It is
closed”, then we do have an error.
Note: There are more interconnections than shown – the End
Time of the ToOpen relation is linked to the Start Time of the
ToBe relation, the logical pins of the ToOpen
and ToBe are linked through an AND.
This looks like too much for a SYNONYMMAP to handle (jumping
from an object to its attribute, handling the case where it is overridden by a later
relation, etc), so we build a ToBe relation, and couple its
start to the finish of ToOpen. When he opens the door, the
door stays open, but can be closed without error.
He opened the door slightly.
This already requires some knowledge of the current state of the door
– if it is half open, it can be opened slightly more. The “slightly”
becomes “slightly” of the remaining interval.
We will assume that “slightly” means 5% of the
interval (whereas “imperceptibly” is relative – if you are going to move
something 20 metres, “imperceptible” isn’t 1% of that, it is, say, 1 mm).
The door was partly open, but too little
for us to enter. We demanded he open the door – he opened it slightly.
We have several sorts of adverbs – one which modifies the action
relation, one which modifies the ToBe relation handling the
state, and one which modifies what was used in the action.
He opened
the door quickly. – acts on the ToOpen relation
He opened the door slightly. – acts
on the ToBe relation
He opened the door carefully. – acts
on the ToUse relation
He opened the door a
little more.
The door was partly open. We asked
him to open it. He closed it, made some adjustment, and then opened it a little more.
Here, we need to go back to a state where the door is open
– we ignore where the door is closed.
We will need to control which ToBe
relation is in control – the latest one. But sometimes we will be hypothesising about
future states – “He may open the door tomorrow” – the action will
occur in the future.
He will open the door halfway
tomorrow, close it that night, then open it a little more the next day.
The statement does not change the current state of the door,
but the future states need to be discoverable from the state of the door, which leads us
to
The order of connections on the Open attribute is irrelevant
– the currently active relation asserts a state, which overrides any existing state
on the attribute.
What do we do with “the door opened”? It seems
reasonable to handle this as a RELATION1 version of “He opened the door”. Then
what about
The door opens onto a courtyard.
This is saying something like “When the door is opened,
access is gained to a courtyard”. It is not saying the door is open at this time.
Hypothetical Situations
We will need a probability accompanying the state, and the ability to
set a hypothetical case true, to see what happens.
The Patient Died
If we use a similar state mechanism for the Alive/Dead state of
a patient:
If the patient dies (from any cause, not just the particular
statement), we get a logical from the ToBe relation to go out
and crush the vital signs and anything else that is relevant. The mechanism could handle
The patient died at 9:03 am, was
resuscitated, but later died at 10:27 am.
There are very few relations that can “undead” a person.
Other State Problems
We have a “common-rare” state for diseases. If we
handle it using state operations, we can accept something being only one in a million (or
even one in a billion – “there are six known cases in the world”), whereas
this precision would get lost in a Bayesian logical value.
The important difference between these states and Bayesian values is
that the spectrum of the state is available wherever it is used, rather than trying to
bundle something into a number and hoping it will be understood elsewhere.
He Broke His Leg
Here we have a change in state which generates a repair
process.
He broke his leg.
His leg is broken.
His leg begins repair.
His leg is mended.
Which looks like (ignoring the fact that
breaking a leg means breaking a bone in the leg)
The creation of the self repair relation is immediate, responding to
the change in state. This seemingly verbose structure allows the repair to take longer
than expected – “The break had to be reset”.
Just as we are shifting the relation parameter from object to
state, it will be necessary to shift from object to component.
“He broke his silence” is a world away from “he
broke his leg”, and we need to catch the difference.
This reduces the number
of possible bones, and allows us to pull up any knowledge about repair problems in those
bones.
States
- a presentation
Relation Hierarchy
Medical Design Notes
