After creating the ontological tree for example after a determined
number of passess of a recursive algorithm finish to cover the tree
and by intuition at least almost each node has a brother we can
rearrange our ontological tree using this kind of inheritance.

Rule 1: all names have probabilty 1 to be a leaf.

Rule 2: if a subset has for the same deep and brotherhood a
corresponding keyword in common we can obtain a normalized version of
the tree putting that element in common to be  child of the parent of
the subset and that subset without the keyword are children of this
keyword. this can be applied to all the tree so we got a normalized
version of the tree in wich leaves are the minimun number of leaves
posible.

Rule 3: the probability to be a element from the same level that "look
like" is not part of a subtree in a parent branch first superset to be
part of the new keyword and subset is sum(children_elements)/
total_number_of_brother_elements, if this probability reaches aprox.
(this number defines how heavy making extra questions is the system).
95% of more for some nodes would be interesting to question again if
they form part of the subset although didn´t appear the relation
"element is a .... keyword".

Regards David.

P.D. if anyone encourages to make the program please upload it , just
for fun to refine the model.

this can make a sugestion of what are the people are searching due to
its properties, (like now does google with the dictionary, google
suggested ...)  for example we can look for a wheeled vehicle in that
case we jus have to look for wheels and vehicle normalized subbranch
this is the result of the jerarquical query all the subbranch elements
are the results of this context or meaning set,so we can make
propositions of what we look for even with that 5% or 10% results that
more or less look like with its properties we can call them the
correction set.the least accesed and queried results can be the
elements that can be removed from the set.this example for organizing
things can be useful for example to huge libraries or organizing
whatever data,give a user and password more or less privileges to
rearrange the sets (that will be number of branches with leaves or
nodes that can be reordered with one question). can also works not
with wikipedia, also can work with a dictionary accepting and
selecting definitions, in this way can be very pompous the system, but
by this way with definitions searching recursevly the meaning of the
words(nouns,the other thread semantically searching) of the definition
we can look for a best match nodes to include the element.

David.

On 9 mayo, 18:31, David_Santana <GAUDELIA1...@yahoo.es> wrote: