The
third stage of the unified categorical Modelling System is where based on the
categorical comprehensive evolutionary/prediction map, the decisions regarding
to the real object are going to be made, when the attributional process in the
second stage by Application, either by an specific application within the
Unified Application itself, or the Unified
Application itself as a global application, is an attributional
processes whose purpose is productive or mixed.
In
fact, there are least two different strategies for the development of the
categorical Modelling System, depending on what kind of categorical
comprehensive model is created in the second stage of the categorical Modelling
System.
The
first strategy, the simplest one perhaps, is the one developed in the first
series of posts dedicated to the specific categorical Modelling System, in the
first phase, that categorical Modelling System as first step in the third stage
of Specific Artificial Intelligences for Productive or Mixed Artificial
Research by Application.
In
the post “Specific categorical Modelling System, second stage”, where I
developed a proposal for the construction of the second stage of the categorical
Modelling System in Specific Artificial Intelligences for Productive or Mixed
Artificial Research by Application, that first proposal for the second stage of
the categorical Modelling System was only: 1) the logical analysis of
categorical sets as first sub-stage of the second stage of the specific
categorical Modelling System (after the first categorical check in the first
stage of the specific categorical Modelling System), analyse internal and
external sets/vectors of the real object (internal are all those vectors in
common with the category attributed, externa all those within the margin of
error so not shared with the category attributed), to avoid contradictions
between sets/vectors, 2) once it has been solved any possible contradiction
between internal/external set/vectors the second sub-stage consists of the
modelling of an static single model, for instance, the model of the farmland in
scale labelling in the farmland what category of lentils was attributed to the
farmland, being in fact a static model to include later on the categorical
comprehensive model of the farm as a static model of the whole farm to include
later on the map, or for instance the model in scale of a package labelling the
category attributed to that package (category according to a possible
combination of variables such as: size, fragility, security, surveillance, risk…),
indicating later on the comprehensive map origin and destiny of the package
along with all the packages participating in the automatic delivery system, 3)
as third sub-stage of the second stage of the categorical Modelling System the
model is set up finally on the map.
Following
this first proposal for the second stage of the specific categorical Modelling
System, then the proposal about how to organize the decision making process, as
I had stated in the post “Specific categorical Modelling System, third stage”, is: according to the label, category
attributed, to the model, category placed in the conceptual scheme attached to
some relevant sets/vectors (for lentils: possible risk of diseases, plague,
need of watering, and any other requirement to make them grow; for the package:
size, fragility, security, surveillance, risk, etc.) the decision making
process consists of the attribution of set of decisions to set of vectors,
decisions about when to plant the lentils, distance between the seeds, how deep
the lentils must be planted on the ground, how often the lentils must be
watered, what pesticides are needed, when to harvest the lentils and how to do
the harvesting, and in the example of the package, decisions about what
packaging is the best option, what means of transport are necessary, what
security and surveillance system is the best option according to the content
and the destination, etc.
In
the first option the categorical single model and the categorical comprehensive
model are static models, and the only role that the category attributed is
going to play on the model is a role as a label indicating over the model the
label of what category has been attributed. Later the single model is included
in the comprehensive model, which is not other thing but a more comprehensive
model in scale where every single object modelled is labelled.
In
fact, the first proposal made for the development of categorical models in the
second stage of the specific categorical Modelling System is no other thing but
the representation in scale of every single object to be included in a more
comprehensive model, where every single model in scale is labelled with the
category attributed, being only a static model.
In
this static model the role played by the category attributed is only as a
label, and the logical analysis of categorical sets made previously in the
first stage of the specific categorical Modelling System when is going to play
a key role is in the third stage of the specific categorical Modelling System
to distribute sets of decisions to sets of vectors, according to external and
internal sets/vectors and according to the position of the single model
labelled in the comprehensive model now inserted on the map, so having in mind
not only categories related to the category of that object, but having in mind
the geographical position and categories and object around the single model of
that object as object of decisions now.
My
first proposal for the construction of the categorical Modelling System the
role played by the categories and the models is very static, I started changing
some elements of this first and earlies proposal in the series of posts
dedicated to the collaboration between by Application and by Deduction between
the deductive and the categorical Modelling System, when I started thinking
about the possibility of evolutionary and predictive models .
And
as long I have been going on working in the categorical Modelling System, by
the time I have started this series of posts dedicated to the unified
categorical Modelling System, and having as a break the last half term, now
what I am developing as a new proposal for the development of the categorical
models to make categorical decisions, in the unified categorical Modelling
System, is a more complex proposal where the categorical models are not static
any longer, the categorical models are definitely dynamic, and the way to make
decisions will change radically, having the categories a more important role
than simply a label in the categorical model, as it was originally in my first
proposal.
The
most important characteristics of the second proposal are:
-
The categorical single and comprehensive models are not static any longer, encompassing
the categorical single evolutionary model, as a dynamic model of the object
according to the internal/external set/vectors attached after the logical
analysis of categorical sets, for instance, a evolutionary model of a land
seeded with lentils, including all kind of details such as how thee seeds
should be planted, watered, fertilized, and use of pesticides, according to the
predictive probabilities associated with every single reaction within the
solution with the highest predictive probability, according to climatic and
geological conditions and chemicals, the chemicals on the ground, the climate,
the lentils, and the expected behaviour of the lentils under such chemical, geological,
climatic conditions, to have a categorical single evolutionary model of how the lentils should be treated (water, use
of fertilizer, pesticides, etc…) and a prediction about how is going to be the
harvest and percentage of productivity. Once the evolutionary and prediction
single categorical models are ready, are integrated evolutionary and prediction
comprehensive categorical model, and later on the evolutionary and prediction
comprehensive map.
-
In this proposal, the role of the categories on the dynamic models is not only
working as labels, but working as a key elements to make prediction single
models, to predict which is going to be the behaviour of a real object under
different scenarios, choosing as most probable that one with the highest
probability. If an farmland is going to seed lentils, in a place like Chile,
the most probable scenario is that sooner or later there is going to be an
earthquake, if the lentils are seeded under tropical weather, the most possible
scenario is that sooner or later are going to suffer a hurricane. Both geological
and climatic phenomena can be predicted when the specific application for the
farm is co-working with other specific applications for climatic or tectonic
within the Unified Application. If an automatic delivery system needs to send a
package to China, and as specific application within the Unified Application is
working with other specific application, such as specific applications for the
National Health System, if the automatic delivery system has to send a package
from a place with an outbreak of high
risk of some diseases, automatically any decision made by the specific
application of the National Health System can be shared with the specific
application for the delivery system, including for those areas with high risk
some new categories related to health and safety in the packaging of any
delivery. In the second proposal the evolutionary models are going to be made
based on the logical analysis of categorical sets, analysing predictive probabilities,
and making evolutionary and prediction models of the real objects based on
predictive probabilities.
- The
predictive probabilities can be calculated having as a source of information,
to calculate the probabilities, the categorical comprehensive evolutionary and
prediction model, according to the expected evolution and prediction for the
categorical comprehensive model, could be possible it could foresee what
probabilities are associated to different phenomena, affecting every single
variable/category/set/vector of any single model, as to make the more realistic
model based on these predictive probabilities.
-
The final aim of making dynamic categorical models based on prediction
probabilities, based on the categorical comprehensive evolutionary and
prediction model, is the possibility that in the future, not very far, the
categorical comprehensive evolutionary model and the prediction model, could be
synthesised with the deductive models of the deductive standardized Modelling
System, as to create one Modelling System, product of the synthesis of the
categorical and deductive single and comprehensive models.
In
essence the reasons for this development from the first proposal for the
categorical Modelling System as to create only static single and comprehensive
models, to this second proposal for the categorical Modelling System more
focused on dynamic categorical single and comprehensive models, are:
-
As soon the first phase for the construction of Specific Artificial Intelligences
for Productive and Mixed Artificial Research by Application is able to evolve
to the second phase, the collaboration between Productive and Mixed Artificial
Research by Application collaborating between them and with Specific Artificial
Intelligences for Artificial Research by Deduction, the higher the
collaboration is between intelligences, sharing more and more information and
categories, the more isomorphic the models can be having access to more
information about different phenomena in different intelligences able to affect
their own models. As long the intelligences can collaborate between them
borrowing information each other, the models that the intelligences could be
able to make are more and more realistic, so that the intelligences could make
more dynamic models involving future events, such as an earthquake or a
hurricane, as to make more realistic predictions about the future to be
included in the single and comprehensive models to locate on the map.
-
Once the second phase gets ready the collaboration between intelligences by
Application as to upgrade to the next phase, in this case the fourth phase, the
unification process, by the time that former specific intelligences by
Application are transformed into specific application within the unified
application having as a sources the unified database of categories to make categorical
attributions, and in the Modelling System the unified conceptual scheme and the
unified categorical comprehensive evolutionary and prediction model, the models
able to make having these sources of information are more and more realistic,
and the reality is that the reality is dynamic changing at any time. If in the
future the idea is to synthesised the categorical comprehensive models and the
deductive comprehensive models, both must be dynamic differentiating
evolutionary and prediction stadiums in their respective representations of the
objects of the real world.
At
some point the evolution from the first proposal for the second and third
stages in the specific categorical Modelling System, models are static and
categories are labels in the model, to the second proposal for the second and
third stages in the unified categorical Modelling System, models are dynamic and
according to the logical analysis of categorical sets in the first stage is
chosen in the second stage the solution as combination of categories with the
highest predictive probability, this evolution is understandable as part of the
logical evolution that should take place from the first experiments in the
first phase, as first period of experimentation in the first phase, to the more
complex experiments, in the first period of experimentation in the fourth
phase.
What
it is foreseeable is that the first experiments in the first phase, in order to
get first prototypes of Specific Artificial Intelligences for Productive or
Mixed Artificial Research by Application, these experiments are going to try to
get ready the first prototypes simplifying as much as possible the modelling process.
Later
on as long the first experiments on the first prototypes of intelligences by
Application are ready to become specific applications within the Unified
Application, along the collaboration process, during this long journey towards
the unification process, as long as the prototypes get more complexity, access
to more sources of information, and the consolidation process is progressing
towards the final model of Global Artificial Intelligence, the first proposal
for the specific categorical Modelling System will let make more complex models
having the second proposal as main method for the creation of categorical
models.
The
evolution that I have shown in my proposal for the organization of the categorical models from the first
phase to the fourth phase, is quite possible that is the evolution that many
categorical models are going to show as long the categorical models evolve from
the specific single and comprehensive categorical models to the unified single
and comprehensive evolutionary and predictive categorical models.
In
this evolution, the way to make decisions there will be an important change
from the first to the fourth phase.
In
the first phase, the third stage of the specific categorical Modelling System
made decisions matching set of decisions to set of vectors, for instance
according to the vector linking the package with the corresponding category of
size, or the corresponding category of destiny to choose the right means of
transport or to set up the risk level or the surveillance, according to these
vectors linking the category of the package with different categories, the
distribution of set of decisions to these set of vectors deciding type of
packaging necessary, means of transport, level of security, surveillance, etc. In essence my first proposal for the third
stage of the specific categorical Modelling System is synthesisable in the idea
that the decision making process is a process of matching set of decisions to
set of vectors, and for that reason previously it is necessary to set up what
sets of decisions are going to be related to what sets of vectors, so that at
any time that a set of vectors is attached to a real object, the third stage
automatically can attribute what set of decisions must be attributed to that
set of vectors, what means that it is necessary to program for ever set of
vectors what set of decisions are related, so that the decision making process
could be defined as an analysis using Venn diagram analysing what set of
decisions converge in every single object analysing what set of vectors
converge in the same real object. The decision making process is done using
Venn diagram.
In
my first proposal for the third stage of the specific categorical Modelling System
in the first phase, the specific decision making process only will require an
analysis by Venn diagram.
In
this first proposal the process could be a little more complex, if in addition
to the analysis using Venn diagram, other variables are introduced such as: the
chronology and temporalization of every decision, for instance, when to plant,
when to water, how often, when to fertilize, how often, when to use pesticides,
how often, and when to harvest. After the application of Venn diagram the
temporalization and frequency of every decision could be programmable, having
as a base for these decisions the result using Venn diagram.
But
in the second proposal the third stage is not so easy, is a bit more complex.
In the second proposal as a base to start the decision making process there is
a dynamic model consisting of the evolutionary and prediction categorical
models already located on the map, and these dynamic model located on the map
is the model of the solution with the highest predictive probability.
The
solution itself is a combination of variables, relevant for the model, related
to a possible set of decisions. In this modelling process before hand is
necessary to program very carefully what conceptual/logical sets/vectors and
what quality set/vectors, as intrinsic variables, of the object itself, are
chosen to be combined with extrinsic variables from the environment, as well as
to program sets of decisions related to every possible combination.
The
categorical model is the model of every single reaction, with very high
predictive probability, due to the solution chosen, solution with the highest
predictive probability, as a product of the combination of intrinsic variables,
from the object itself (conceptual/logical and quality set/vectors) and
extrinsic variables, from the environment, and possible set of decisions for
every combination.
In
the modelling process the way to make combinations between intrinsic variables (from
the object itself) and extrinsic variables (from the environment), is to mingle
all possible discrete categories related to any conceptual/logical or quality
set/vector from the object with all possible discrete category from the environment,
analysing every single combination as a possible combination related to a
possible solution as a set of decisions, matching possible sets of decisions
for every solution, and calculating the predictive probability of every
solution and decision, and for every solution and decision analysing every
possible reaction, calculating the probability of every possible reaction.
In
the end the model is the model of the solution/decision with the highest
predictive combination and possible reactions with the highest predictive
probability, of that combination of intrinsic and extrinsic variables,
intrinsic from the object and extrinsic from the environment, more likely to happen
attached to the corresponding set of decisions. Placing the model of the
solution/decision more likely to happen as a dynamic model including
foreseeable evolution and foreseeable prediction within the categorical evolutionary
and prediction model, to be placed finally on a categorical dynamic map.
In
brief, the methodology of the second proposal for dynamic categorical models
is:
-
Identification of relevant conceptual/logical and quality set/vectors of the
object as intrinsic variables of the object, according to the categorical
attribution in the second stage by Application and how it was filed and
analysed in the conceptual scheme as first stage of the Modelling System.
-
Identification of relevant variables from the environment, identifiable within
the categorical comprehensive evolutionary model.
- Combination
of all intrinsic and extrinsic variables between them.
-
Calculation of predictive probabilities for every combination.
-
Analysis of possible reactions to every combination.
-
Calculation of predictive probabilities for every reaction.
-
Attribution of set of decisions to the most likely combination and reactions,
as a result, this should be the most likely solution.
But
this dynamic model is only a model although dynamic. In this dynamic model is
not represented only the farmland in scale, or the package in scale, but the
model of everything, is the whole model
of the solution, what means, the model of the chosen decision according to what
set of decisions is related to the solution with the highest probability,
modelling: from how to plant the seeds (distance between seeds, and depth), how
to water and how often, how to fertilize and how often, when pesticides are
necessary and how often, when and how to harvest, or what packaging is necessary,
what means of transport are necessary, security level and surveillance systems,
modelling the chronology of every single event during the plantation and the
delivery of the product.
Saying
that this model is only a dynamic model of the solution/decision what I say is
that till now this is only a model, till now, no decision has been ordered to
the Decision System, till now, the only thing that the specific application has
done is only a simple drawing of a model of one solution attached to a set of
decisions to put on a map, nothing else, till now, the only thing that the
specific application has is only a very realistic and detailed representation
of something real in scale, but till now there has not been ordered any
decision to the Decision System, it is only a drawing.
Till
now what the second stage of the unified categorical Modelling System has done,
is using Venn diagram the specific application has identified intrinsic
variables (sets/vectors analysed on the conceptual scheme, sets/vectors related
to lentils or the package) in the object itself, and from the categorical
comprehensive model what extrinsic variables from the environment can interact
in the process, and using combinatory, to make as many combinations as possible
between intrinsic and extrinsic discrete categories associated with intrinsic
and extrinsic variables, calculating predictive probabilities for every
combination/solution related to a set of decisions, having as a source for this
calculation the comprehensive evolutionary and prediction model/map.
What
the specific application for the farm or the automatic delivery system have
within the Unified Application once the second stage of the unified categorical
Modelling System is completed is a very realistic model of the more likely
model on the map.
Once
the more probable model on the map passes to the third stage, is when it is
time for the decision making process, analysing the model on the map.
The
model on the map is no other thing but a very realistic drawing of a possible solution as a set of possible
decisions related to some combination of intrinsic and extrinsic categories.
Every
time we look up on google maps, what we are obtaining is a list of possible
solutions to our future journey, from one point to another, where every possible
combination of categories, car, uber, taxi, walking or public transport, and
within public transport the bus, the train, the tube, the over-ground, is a
possible solution attached to the expected time of travelling, what it is going
to be the most important criterion in our election of what combinatory is more
suitable for our journey.
What
we got on google maps is only a drawing, no decision has been made. Once we
check the possible solutions according to the different combinations, and
attached to every combination how long it would take the journey we make a
decision.
In
this case the decision is more likely dependant on how long does it take, but
there are other criteria, such as, criteria related to our personality, like
what we prefer, over ground to have a view of the city of London or tube to go
faster, or bus because is cheaper.
In
the same way that we make the decision on google maps depending on our
personality or criteria like time, or speed, the categorical comprehensive
evolutionary map will offer the most likely solution, and once the more likely
solution is determined, according to the solution, the third stage of the
unified categorical Modelling System is going to transform the set of decisions
attributed to the most likely combination in the solution, as a set of
decisions to be ordered to the Decisional System.
At
the end, in the second proposal, the solution is the decision, but the decision
is not a real decision until it has been included in the categorical
comprehensive evolutionary and prediction model, because, if the solution has
any contradiction with any other single model in the comprehensive model, is
the fourth categorical check in the comprehensive evolutionary and prediction
map where this contradiction should be solved, so that the set of decisions
involved in the solution can be transformed by the third stage of the
categorical Modelling System into a set of decisions to be filed in the
database of decisions as first stage of the categorical Decisional System
ensuring that at least on the dynamic categorical map it had not any further
contradiction, and analysing in the fifth categorical check in the third stage
of the unified categorical Modelling System the absence of contradictions
between the decisions involved in the set of decisions attached to the solution
of the more likely combination of intrinsic and extrinsic variables.
In
the end the mathematics behind the second proposal for the decision making
process in the unified categorical Modelling System is not so different to
other proposals made along this blog, being a result of: combinatory (of
variables), probability (what combination is more likely), Venn diagram (matching
set of decisions to the most likely combination of variables).
The
mathematic behind the decision making process is not so different to Yolanda,
with the difference that instead of empirical probability, the categorical
decisions are based on predictive probability, but this is only my proposal, in
the proposal that other intelligence agencies are going to carry out are going
to combine empirical probability and prediction probability, for instance, once
it has been calculated the predictive probability of every solution, according
to the records, which is the empirical probability of every possible reaction.
In
Yolanda in 2003 I did not speak about Venn Diagram, but the wardrobe of Yolanda
could be considered as a possible set of decisions for possible environmental variables,
in the end is the same, if it is raining Yolanda will take the umbrella, what
is no other thing that a set of decisions has been attached to a set of
variables, this is Venn diagram. And the
way in which Yolanda can choose some outfit: sneakers, jeans and t-shirt, instead
of black high heels, blue skirt and white shirt, is no other thing that a
possible combination of clothes related to some activity.
The
decision making process could be synthesised as a process where using empirical
or predictive probabilities according to combinatory of variables is possible
by Venn diagram to match decisions.
What
in the second proposal for the unified categorical Modelling System is very
important is to keep a permanent surveillance along all the process, due to the
dynamicity of the system. Because the system is going to be based on dynamic
models and dynamic maps, this means that at any time that there is a change in
the dynamic models and map, there should be changes in all the single models
included in the dynamic model and map susceptible to suffer variations due to
variations of other single models.
If
a single model of how to send a package from Los Angeles to London suddenly has
to face a hurricane coming from Cuba, to the north Atlantic, the categorical
model and map should make changes in order to face changes in the delivery of the
package. At any time that the categorical comprehensive evolutionary or
prediction model has a change, for any reason, the third categorical check
should be able to identify any possible contradiction to make changes in the
structure of the current single models on the comprehensive model, changes that
should be communicated to the comprehensive map, rearranging the current
decisions on the third stage making as many changes as necessary as update of
the decisions in the third stage, update to be assessed by the fifth
categorical check in the third stage of the Modelling System, to be communicated
to the database of decision in the categorical Decisional System.
The
way in which the solution with the highest predictive probability once it has
been placed on the map not having contradictions at all, will be transformed
into a range of decisions, is transforming every single decision of the set of
decisions attached to the combination of variables as a decision to be labelled
with sub-factor (position), sub-section (subject), priority, and time.
The
priority of every decision in the third stage should be calculated using for
instance Impact of the Defect in case that it is not applied and/or Effective
Distribution, the importance of that decision in the chain of decisions to
achieve high effectivity.
The
time of every decisions should be calculated using estimations of time, when a
decision is expected, and making as many arrangements depending on the matter
according to the intrinsic and extrinsic variables, for instance, when to
plant, when to water and how often, when to fertilize and how often, when to
use pesticides and how often, and when to harvest.
Here
the Impact of the Defect or the Effective Distribution will play an important
role, because maybe if the plantation is not watered one time, is not going to
have a great impact, if it is not watered two times, could have some impact,
but not important, but if it is not watered three or more times the impact of
the product could be very important. Time and priority quite possible are going
to be very related.
The
tasks to perform the third stage of the unified categorical Modelling System
could be stated as follows:
-
The third stage of the unified categorical Modelling System must transform the
solution in a range of decisions to be ordered to the Decisional System.
-
In order to get ready the range of decisions to be ordered to the Decisional
System, the decisions must pass the fifth categorical check to ensure the
absence of contradiction between decisions.
-
Calculation of the priority level of every decision, using for that purpose
techniques such as the Impact of the Defect and/or the Effective Distribution.
-
Calculation of when every decision should be applied, time
-
Analysis of what sub-factor (position) corresponds to every decision.
-
What is the sub-section (subject) of every decision.
-
The third stage of the unified categorical Modelling System files every
decision in the corresponding place in the database of decisions, as a Russian
Dolls system of categorical decisions, as first stage of the categorical
Decisional System, where the first stage of the categorical Decisional System
will make a quick check or the first categorical adjustment, depending on the
priority level of every decision, to ensure the absence of contradiction
between every new decision and the decisions already on the project, and in
case of contradictions, any rearrangement of decisions will be made according
to the adaptation rule, what means, the less priority decision must be adapted
to the more priority decisions, in case of partial contradictions, because in
case of total contradictions, the less priority decision will be sent back the
source, in this case the less priority decision will be sent back again to the
categorical Modelling System to be rearranged again without contradiction.
In essence,
the development for the third stage by application for productive or mixed intelligences
is not different that that one working by Deduction, with the only difference
that now the decisions are not based on rational decisions, equations, but on
categorical attributions, so all the decisional process is categorical,
matching sets of decisions to set of categories relate to the categorical
attribution, and later on, the process is not so different, calculating
priority levels, ordering according to position and subject, sub-factor and
sub-section, and priority levels, and when the decision should be applied
